Word2Vec_using_MLP.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/mdaftary/Word2Vec-Algo/blob/main/Word2Vec_using_MLP.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "g-W1u8iQ4DqL"
      },
      "source": [
        "\n",
        "\n",
        "\n",
        "![Word2Vec](https://serokell.io/files/qz/qzp39hlo.pic_7_Word2Vec.png)\n",
        "Image credits: https://serokell.io/blog/word2vec"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ivgkA-Gj7JjW"
      },
      "source": [
        "# **Section 1** Introduction and motivation to Word2Vec (no tasks to complete)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ucepd5-c8vOD"
      },
      "source": [
        "Have you ever wondered how large language models like ChatGPT and BARD process text?\n",
        "In this programming assignment, we are going to explore Word2Vec, a popular algorithm used in natural language processing (machine learning on text) for models to \"understand\" the human language.  We will use MLP to build our very own Word2Vec model.\n",
        "\n",
        "The need for these algorithms comes from the fact that machine learning models operate on numbers but not text.  Because of this, before you train your text model, you first need a method to map your text into numbers.\n",
        "\n",
        "Perhaps the most naive way to do so is by assigning a random number to each word.  For example, let's say we have the following text (taken from Dr Suess, Green Eggs and Ham):\n",
        "\n",
        "<center>\n",
        "\n",
        "> *I do not like them here or there.* <br>\n",
        "> *I do not like them anywhere.* <br>\n",
        "> *I do not like green eggs and ham.*\n",
        "</center>\n",
        "\n",
        "We *could* just assign numbers randomly, and one such mapping could be\n",
        "\n",
        "```python\n",
        "word_to_index = {\n",
        "  'I': 0,     'them': 1,     'here': 2,    ...\n",
        "  'there': 9, 'eggs': 10,    'do': 11      ...\n",
        "  'ham': 16\n",
        "}\n",
        "```\n",
        "\n",
        "but the problem with this approach is that it does not encode *any* information about the meanings of the words at all.  For example, 'eggs' and 'ham' are similar words (both are food), but have vastly different values (eggs is 10, ham is 16).  \n",
        "\n",
        "To illustrate the problem with this, let's say you give your model the word 'green', in the hope that it predicts the next word to be 'eggs'.  After training for a few epochs, it predicts the next word to be 'I', which is obviously wrong.  You train your model for a few more epochs and re-feed the word 'green'.  Now it predicts the next word to be 'ham'.  You want to tell the model that it did a better job than before, but since the numbers associated with each word are assigned randomly, you cannot use the difference between the numbers to tell your model that it got closer to the true answer.  In other words, you are unable to tell your model anything other than \"you got it right\" or \"you got it wrong\".  This makes training your model much harder.\n",
        "\n",
        "A good mapping should, therefore, map similar words close to each other, which is the goal of Word2Vec.  The following *2D semantic space* shows an example of what we want to achieve.\n",
        "\n",
        "*(If the image does not load, run the code cell for \"2D Sematic Space\" in the appendix of the notebook to generate the image)*\n",
        "\n",
        "![img](https://course.cse.ust.hk/comp2211/assignments/pa2/images/2d_semantic_space.png)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "O2tNZG-R-7qZ"
      },
      "source": [
        "Using the two *semantic features* \"animal\" and \"food\", we are able to encode the words above in a much more meaningful way.  Each word is now represented by a vector in $\\mathbb{R}^2$ (a list of 2 real numbers) and positioned such that similar words are clustered together:\n",
        "\n",
        "<center>\n",
        "\n",
        "|           | food (x-axis) | animal (y-axis) |\n",
        "|-----------|:-------------:|:---------------:|\n",
        "| llama     |      0.25     |       5.25      |\n",
        "| dog       |      0.5      |        5        |\n",
        "| frog      |      0.75     |       4.75      |\n",
        "| coriander |       5       |        1        |\n",
        "| ...       |      ...      |       ...       |\n",
        "\n",
        "</center>\n",
        "\n",
        "**Question**: Where do you think the words \"rat\", \"lamb\", and \"grapes\" would be placed?\n",
        "\n",
        "This is already a great step up from giving each word random values.  But why stop at just 2 semantic features?  Let's say we want to add the words \"cat\", \"rabbit\" and \"hamster\" to the semantic space.  They would probably be placed in the top-left corner, together with \"llama\", \"dog\", and \"frog\".  But obviously, cats, dogs, rabbits and hamsters have something else in common: they are all pets!  So, it would make sense to add a \"pet\" semantic feature, enriching our semantic space.  The result may look something like the following picture:\n",
        "\n",
        "*(If the image does not load, run the code cell for \"3D Sematic Space\" in the appendix of the notebook to generate the image)*\n",
        "\n",
        "![img](https://course.cse.ust.hk/comp2211/assignments/pa2/images/3d_semantic_space.png)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ywa1FKIjOpue"
      },
      "source": [
        "Now, we have encoded each word as a vector in $\\mathbb{R}^3$ (a list of 3 real numbers):\n",
        "\n",
        "<center>\n",
        "\n",
        "|           | food (x-axis) | animal (y-axis) | pet (z-axis) |\n",
        "|-----------|:-------------:|:---------------:|:------------:|\n",
        "| llama     |      0.25     |       5.25      | 0 |\n",
        "| dog       |      0.5      |        5        | 5 |\n",
        "| frog      |      0.75     |       4.75      | 1 |\n",
        "| coriander |       5       |        1        | 0 |\n",
        "| ...       |      ...      |       ...       |...|\n",
        "\n",
        "</center>\n",
        "\n",
        "But there is no reason to stop at 3 semantic features.  Evidently, by continuously adding even more semantic features (dimensions), our Word2Vec model can become even more expressive and capable of capturing the different intertwined meanings of our vocabulary.  \n",
        "\n",
        "Typically, large language models use Word2Vec models with 300 dimensions - a list of 300 real numbers for *each* vocabulary!  What's more, Word2Vec models are trained in an unsupervised manner.  This means that we do not supply the labels (i.e. \"food\", \"animal\", \"pet\") to the dimensions.\n",
        " Instead, we ask the model to learn them itself."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "J1sG3AGTsfRh"
      },
      "source": [
        "There are 2 methods we can use to learn the dimensions: the **continuous bag of words** (CBOW) approach and the **skip-gram** approach.  The idea of both approaches is to train the model by feeding it a huge amount of text (for example, all Wikipedia articles), but they differ in how we represent the context words.  Let's illustrate the differences by considering the following sentence:\n",
        "\n",
        "> The Sydney Opera House is a multi-venue performing arts centre in Sydney.\n",
        "\n",
        "In the **continuous bag of words** approach, **the goal is to predict *one* target word from a few context words**.  So, for each target word in the sentence, we look up *n* context words (in this programming assignment, we will take *n=2*) to the left and right of the target word, and construct a context table, as follows:\n",
        "\n",
        "*Note:  the size of n is formally called the \"window size\".*\n",
        "\n",
        "<center>\n",
        "\n",
        "| target word (y) | context words (x)         |\n",
        "|:-----------:|------------------------|\n",
        "|     the     | sydney, opera          |\n",
        "|    sydney   | the, opera, house      |\n",
        "|    opera    | the, sydney, house, is |\n",
        "|    house    | sydney, opera, is, a   |\n",
        "|      is     | opera, house, a, multi |\n",
        "|     ...     | ...                    |\n",
        "\n",
        "</center>\n",
        "\n",
        "In the **skip-gram** approach, **the goal is to predict a few context words *given* one target word**.  So, for each target word in the sentence, we also look up *n* context words (*n=2*), but each context word is recorded separately, as follows:\n",
        "\n",
        "<center>\n",
        "\n",
        "| target word (x) | context words (y)|\n",
        "|:-----------:|:-------------:|\n",
        "|     the     |     sydney    |\n",
        "|     the     |     opera     |\n",
        "|    sydney   |      the      |\n",
        "|    sydney   |     opera     |\n",
        "|    sydney   |     house     |\n",
        "|    opera    |      the      |\n",
        "|    opera    |     sydney    |\n",
        "|    opera    |     house     |\n",
        "|    opera    |       is      |\n",
        "|    house    |     sydney    |\n",
        "|     ...     |      ...      |\n",
        "\n",
        "</center>\n",
        "\n",
        "In the continuous bag of words approach, we feed the model the context words and ask it to predict one target word.  In the skip-gram approach, we feed the model one target word and ask it to predict context words.\n",
        "\n",
        "The model architecture is as follows:\n",
        "\n",
        "$$\n",
        "\\begin{align*}\n",
        "y &= \\text{softmax}\\big(T_2 (T_1 x)\\big), \\hspace{1em}\\text{where} \\\\\n",
        "x &\\text{ is a one-hot encoded vector in } \\mathbb{R}^{\\text{num_vocabs}} \\\\\n",
        "T_1 &\\text{ is a linear transformation from } \\mathbb{R}^{\\text{num_vocabs}}\\to\\mathbb{R}^{300} \\\\\n",
        "T_2 &\\text{ is a linear transformation from } \\mathbb{R}^{300}\\to\\mathbb{R}^{\\text{num_vocabs}} \\\\\n",
        "y &\\text{ is a vector in }\\mathbb{R}^{\\text{num_vocabs}}\n",
        "\\end{align*}\n",
        "$$\n",
        "\n",
        "*(The appendix of this notebook has a quick review on one-hot encoding)*\n",
        "\n",
        "Put simply, the model consists of 2 layers.\n",
        "- The first layer takes each word and outputs a vector in $\\mathbb{R}^{300}$.\n",
        "- The second layer takes these vectors and maps them back down to $\\mathbb{R}^{\\text{num_vocabs}}$.  The output is run through a softmax function to determine which word the model predicts.\n",
        "\n",
        "Note that the model architecture doesn't change no matter if we use the continuous bag of words or skip-gram approach, the only difference lies in how we encode the input words.\n",
        "\n",
        "At this point, you may wonder how predicting context words can help us \"group similar words together\".  Notice that our ultimate goal is to find a function that maps words to vectors of 300 dimensions and not predict context words, and in fact, our $T_1$ linear layer does exactly that:  it takes in a one-hot vector representing the word and outputs a vector in 300 dimensions.\n",
        "\n",
        "\n",
        "In the following tasks, we will build our own Word2Vec model.  For easier debugging and to test if the model was built correctly, we will train our model using a short paragraph instead of the whole Wikipedia articles dataset.  The short paragraph chosen is taken from the Wikipedia page on [*open sets*](https://en.wikipedia.org/wiki/Open_set) in mathematics."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pRZC_bm27WMy"
      },
      "source": [
        "# **Section 2** Generating a simple dataset for training (3 tasks to complete)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "u4cpIWZ9rpsh"
      },
      "source": [
        "The first step is cleaning the text by removing punctuation marks and capitalising.  This step has been done for you."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "R0p_a6_4p0OO",
        "outputId": "6c8f4e29-2cb9-46f4-c50c-cba36018f923"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "in mathematics an open set is a generalization of an open interval in the real line in a metric space a set along with a distance defined between any two points an open set is a set that along with every point p contains all points that are sufficiently near to p that is all points whose distance to p is less than some value depending on p more generally an open set is a member of a given collection of subsets of a given set a collection that has the property of containing every union of its members every finite intersection of its members the empty set and the whole set itself a set in which such a collection is given is called a topological space and the collection is called a topology these conditions are very loose and allow enormous flexibility in the choice of open sets for example every subset can be open the discrete topology or no subset can be open except the space itself and the empty set the indiscrete topology in practice however open sets are usually chosen to provide a notion of nearness that is similar to that of metric spaces without having a notion of distance defined in particular a topology allows defining properties such as continuity connectedness and compactness which were originally defined by means of a distance\n"
          ]
        }
      ],
      "source": [
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "#-# STEP 1:  Clean text.  Remove punctuations, caps #-#\n",
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "\n",
        "# do not edit the text below or the cleaning may not work!\n",
        "if __name__ == '__main__':\n",
        "\n",
        "  text = '''In mathematics, an open set is a generalization of an open interval in the real line.\n",
        "  In a metric space (a set along with a distance defined between any two points),\n",
        "  an open set is a set that, along with every point P, contains all points that are sufficiently near to P\n",
        "  (that is, all points whose distance to P is less than some value depending on P).\n",
        "  More generally, an open set is a member of a given collection of subsets of a given set,\n",
        "  a collection that has the property of containing every union of its members, every finite intersection of its members,\n",
        "  the empty set, and the whole set itself. A set in which such a collection is given is called a topological space,\n",
        "  and the collection is called a topology. These conditions are very loose, and allow enormous flexibility in the choice of open sets.\n",
        "  For example, every subset can be open (the discrete topology), or no subset can be open except the space itself and the empty set\n",
        "  (the indiscrete topology). In practice, however, open sets are usually chosen to provide a notion of nearness that is\n",
        "  similar to that of metric spaces, without having a notion of distance defined. In particular,\n",
        "  a topology allows defining properties such as continuity, connectedness, and compactness, which were originally defined by means of a distance.'''\n",
        "\n",
        "  import re  # python's regular expression library.  A powerful library for text manipulation\n",
        "  cleaning = text.lower()  # CAPS -> caps\n",
        "  cleaning = re.sub(r'\\n', ' ', cleaning)  # replace the newline characters with a space\n",
        "  cleaning = re.sub(r'[^a-z0-9 -]', '', cleaning)  # replace all characters except letters, numbers, space bars and hyphens with an empty string\n",
        "  cleaned_text = re.sub(' +', ' ', cleaning)  # replace multiple spaces with one space\n",
        "\n",
        "  print(cleaned_text)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "grM--opddyQX"
      },
      "source": [
        "Next, use the cleaned text to create the following **four** variables:\n",
        "\n",
        "<center>\n",
        "\n",
        "| variable name  | datatype       | description                                         |\n",
        "|----------------|----------------|-----------------------------------------------------|\n",
        "|`words_in_order`|`list[str]`     | a list of all the words (which are separated by spaces) in the order they appeared in `cleaned`   |\n",
        "|`num_vocabs`    |`int`           | the number of unique words (i.e. vocabs)            |\n",
        "|`word_to_index` |`dict[str, int]`| a dictionary mapping each unique word to an integer.  The mapping should follow alphabetic order |\n",
        "|`index_to_word` |`dict[int, str]`| a dictionary mapping each integer back to the word  |\n",
        "\n",
        "</center>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WP-QuZBfiK3J"
      },
      "source": [
        "*(This cell can be deleted)*  For example, for the cleaned text\n",
        "> the sydney opera house is a multi-venue performing arts centre in sydney\n",
        "\n",
        "the four variables may look as follows:\n",
        "\n",
        "<center>\n",
        "\n",
        "| variable name | content                                                                                                                               |\n",
        "|---------------|---------------------------------------------------------------------------------------------------------------------------------------|\n",
        "|`words_in_order`|`['the', 'sydney', 'opera', 'house', 'is', 'a', 'multi-venue', 'performing', 'arts', 'centre', 'in', 'sydney']`|\n",
        "|`num_vocabs`   |`11`                                                                                                                                   |\n",
        "|`word_to_index`|`{'a': 0, 'arts': 1, 'centre': 2, 'house': 3, 'in': 4, 'is': 5, 'multi-venue': 6, 'opera': 7, 'performing': 8, 'sydney': 9, 'the': 10}`|\n",
        "|`index_to_word`|`{0: 'a', 1: 'arts', 2: 'centre', 3: 'house', 4: 'in', 5: 'is', 6: 'multi-venue', 7: 'opera', 8: 'performing', 9: 'sydney', 10: 'the'}`|\n",
        "\n",
        "</center>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "PfwcSQeFaovH"
      },
      "outputs": [],
      "source": [
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "#-# STEP 2:  Compile corpus, create word_to_index and index_to_word #-#\n",
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "\n",
        "def generate_variables(cleaned_text):\n",
        "  \"\"\"\n",
        "  Generates the four required variables above.\n",
        "  Args:\n",
        "    cleaned_text: the cleaned text in the step above\n",
        "  Returns:\n",
        "    words_in_order, num_vocabs, word_to_index, index_to_word\n",
        "  \"\"\"\n",
        "\n",
        "  ### TO DO ###\n",
        "  words_in_order = cleaned_text.split()\n",
        "  words_set = sorted(set(words_in_order))\n",
        "  num_vocabs = len(words_set)\n",
        "  word_to_index = dict(map(reversed, enumerate(words_set)))\n",
        "  index_to_word = dict(enumerate(words_set))\n",
        "\n",
        "  return words_in_order,num_vocabs,word_to_index,index_to_word\n",
        "\n",
        "if __name__ == '__main__':\n",
        "  words_in_order, num_vocabs, word_to_index, index_to_word = generate_variables(cleaned_text)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1sqH3phQkqTF"
      },
      "source": [
        "Next, generate datasets for both the skip-gram approach and the continuous bag of words approach.  Complete the following **two** functions,\n",
        "\n",
        "<center>\n",
        "\n",
        "| function name | parameter name | parameter datatype | return type\n",
        "|---------------------------------|------------------------------------------|----------------------------|-------------------------------|\n",
        "| `generate_training_data_skipgram` | `words_in_order`<br> `word_to_index`<br> `window_size` | `ls[str]`<br> `dict[str, int]`<br> `int` | `tuple(np.ndarray, np.ndarray)` |            |\n",
        "| `generate_training_data_cbow`     | `words_in_order`<br> `word_to_index`<br> `window_size` | `ls[str]`<br> `dict[str, int]`<br> `int` | `tuple(np.ndarray, np.ndarray)` |\n",
        "\n",
        "</center>\n",
        "\n",
        "and create the following **four** variables:\n",
        "\n",
        "<center>\n",
        "\n",
        "| variable name | datatype     | description                                                                              |\n",
        "|---------------|--------------|------------------------------------------------------------------------------------------|\n",
        "| `x_cbow`      | `np.ndarray` | 2D array of the cbow **context** word table, where each word is represented by its index.  Use `window_size = 2`      |\n",
        "| `y_cbow`      | `np.ndarray` | 1D array of the cbow **target** word table, where each word is represented by its index.  Use `window_size = 2`     |\n",
        "| `x_skipgram`  | `np.ndarray` | 1D array of the skipgram **target** word table, where each word is represented by its index.  Use `window_size = 2`  |\n",
        "| `y_skipgram`  | `np.ndarray` | 1D array of the skipgram **context** word table, where each word is represented by its index.  Use `window_size = 2` |\n",
        "\n",
        "</center>\n",
        "\n",
        "Note that in case the target word has fewer context words than usual `window_size*2` (e.g. the first, second, second last, and last words with respect to `window_size=2`), you need to 'pad' the context word lists by adding -1 at the end."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SGC8uLh6XyqS"
      },
      "source": [
        "*(This cell can be deleted)*  For example, using the results from STEP 2, the four variables may look as follows:\n",
        "\n",
        "<center>\n",
        "\n",
        "| variable name | content                                                                                                                                                                       |\n",
        "|---------------|-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|\n",
        "| `x_cbow`      | `[[ 9  7 -1 -1] [10  7  3 -1] [10  9  3  5] [ 9  7  5  0]  [ 7  3  0  6] [ 3  5  6  8] [ 5  0  8  1] [ 0  6  1  2]  [ 6  8  2  4] [ 8  1  4  9] [ 1  2  9 -1] [ 2  4 -1 -1]]` |\n",
        "| `y_cbow`      | `[10  9  7  3  5  0  6  8  1  2  4  9]`                                                                                                                                       |\n",
        "| `x_skipgram`  | `[10 10  9  9  9  7  7  7  7  3  3  3  3  5  5  5  5  0  0  0  0     6  6  6  6  8  8  8  8  1  1  1  1  2  2  2  2  4  4  4  9  9]`                                          |\n",
        "| `y_skipgram`  | `[ 9  7 10  7  3 10  9  3  5  9  7  5  0  7  3  0  6  3  5  6  8     5  0  8  1  0  6  1  2  6  8  2  4  8  1  4  9  1  2  9  2  4]`                                          |\n",
        "\n",
        "</center>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "pJfMADBtamk0"
      },
      "outputs": [],
      "source": [
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "#-# STEP 3:  Generate training data part 1  #-#\n",
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "\n",
        "import numpy as np\n",
        "\n",
        "def generate_training_data_cbow(words_in_order, word_to_index, window_size):\n",
        "    \"\"\"\n",
        "    Generates the target word table and context word table based on the continuous bag of words approach.\n",
        "    The table should contain the index of the word instead of the word itself.\n",
        "    For words in the start and end of the whole text where the number of context words are less than window_size * 2,\n",
        "    fill up the remaining entries of the array with 'special number' -1 instead\n",
        "    Args:\n",
        "      words_in_order: the list of words appearing in order that we created in the previous step\n",
        "      word_to_index: the dictionary that we created in the previous step\n",
        "      window_size: how far we scan for context words\n",
        "    Returns:\n",
        "      a tuple of 2 numpy arrays\n",
        "    \"\"\"\n",
        "    ### TO DO ###\n",
        "    len_words = len(words_in_order)\n",
        "    cbow_data = np.full((len_words,window_size*2), -1)\n",
        "    y_cbow = np.empty((len_words,),dtype = int)\n",
        "    for i , data in enumerate(words_in_order):\n",
        "      y_cbow[i] = word_to_index[data]\n",
        "      count = 0\n",
        "      for back in range(min((i,window_size)),0,-1):\n",
        "        cbow_data[i,count] = word_to_index[words_in_order[i-back]]\n",
        "        count += 1\n",
        "      for front in range(min((window_size,len_words-i-1))):\n",
        "        cbow_data[i,count] = word_to_index[words_in_order[i+front+1]]\n",
        "        count += 1\n",
        "    return (cbow_data,y_cbow)\n",
        "\n",
        "\n",
        "def generate_training_data_skipgram(words_in_order, word_to_index, window_size):\n",
        "    \"\"\"\n",
        "    Generates the target word table and context word table based on the skipgram approach.\n",
        "    The table should contain the index of the word instead of the word itself.\\\n",
        "    Args:\n",
        "      words_in_order: the list of words appearing in order that we created in the previous step\n",
        "      word_to_index: the dictionary that we created in the previous step\n",
        "      window_size: how far we scan for context words\n",
        "    Returns:\n",
        "      a tuple of 2 numpy arrays\n",
        "    \"\"\"\n",
        "    ### TO DO ###\n",
        "    len_words = len(words_in_order)\n",
        "    x_skipgram = []\n",
        "    y_skipgram = []\n",
        "    for i, data in enumerate(words_in_order):\n",
        "      for back in range(min((i,window_size)),0,-1):\n",
        "        x_skipgram.append(word_to_index[data])\n",
        "        y_skipgram.append(word_to_index[words_in_order[i-back]])\n",
        "      for front in range(min((window_size,len_words-i-1))):\n",
        "        x_skipgram.append(word_to_index[data])\n",
        "        y_skipgram.append(word_to_index[words_in_order[i+front+1]])\n",
        "    return (np.array(x_skipgram),np.array(y_skipgram))\n",
        "\n",
        "if __name__ == '__main__':\n",
        "  x_cbow, y_cbow = generate_training_data_cbow(words_in_order, word_to_index, 2)\n",
        "  x_skipgram, y_skipgram = generate_training_data_skipgram(words_in_order, word_to_index, 2)\n",
        "  '''\n",
        "  print(x_cbow)\n",
        "  print(x_cbow.shape)\n",
        "  print(y_cbow)\n",
        "  print(y_cbow.shape)\n",
        "  print(x_skipgram)\n",
        "  print(x_skipgram.shape)\n",
        "  print(y_skipgram)\n",
        "  print(y_skipgram.shape)\n",
        "  '''"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LV-VX5800ZCN"
      },
      "source": [
        "Finally, convert the variables `x_skipgram`, `y_skipgram`, `x_cbow` and `y_cbow` created in the previous step to their one-hot representations. Complete the following **two** functions,\n",
        "\n",
        "<center>\n",
        "\n",
        "| function name              | parameter name        | parameter datatype    | return type  |\n",
        "|----------------------------|-----------------------|-----------------------|--------------|\n",
        "| `one_hot`                  | `arr`<br>`num_vocabs` | `np.ndarray`<br>`int` | `np.ndarray` |\n",
        "| `one_hot_multicategorical` | `arr`<br>`num_vocabs` | `np.ndarray`<br>`int` | `np.ndarray` |\n",
        "\n",
        "</center>\n",
        "\n",
        "and create the following **four** variables:\n",
        "\n",
        "<center>\n",
        "\n",
        "| variable name     | datatype     | description                                                      |\n",
        "|-------------------|--------------|------------------------------------------------------------------|\n",
        "| `x_cbow_1hot`     | `np.ndarray` | a 2D array containing the multicategorical one-hot representation of `x_cbow`     |\n",
        "| `y_cbow_1hot`     | `np.ndarray` | a 2D array containing the one-hot representation of `y_cbow`     |\n",
        "| `x_skipgram_1hot` | `np.ndarray` | a 2D array containing the one-hot representation of `x_skipgram` |\n",
        "| `y_skipgram_1hot` | `np.ndarray` | a 2D array containing the one-hot representation of `y_skipgram` |\n",
        "\n",
        "</center>\n",
        "\n",
        "*Note:  it is possible to complete both functions without any loops or comprehensions.* **To score full marks for this task, you must not use explicit loops, comprehensions, or import libraries to help you solve this task.  In other words, solve it using purely vectorization techniques.**  Refer to the grading scheme on this programming assignment's webpage for details on the scoring."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VrdKbqWGLSGC"
      },
      "source": [
        "*(This cell can be deleted)*  For example, using the results from STEP 3, the four variables may look as follows:\n",
        "\n",
        "<center>\n",
        "\n",
        "| variable name     | content                                                                                                                      |\n",
        "|-------------------|------------------------------------------------------------------------------------------------------------------------------|\n",
        "| `x_cbow_1hot`     | `[[0 0 0 0 0 0 0 1 0 1 0]`<br> ` [0 0 0 1 0 0 0 1 0 0 1]`<br> ` [0 0 0 1 0 1 0 0 0 1 1]`<br>  ` ...`<br> ` [0 0 1 0 1 0 0 0 0 0 0]]`     |\n",
        "| `y_cbow_1hot`     | `[[0 0 0 0 0 0 0 0 0 0 1]`<br> ` [0 0 0 0 0 0 0 0 0 1 0]`<br> ` [0 0 0 0 0 0 0 1 0 0 0]`<br>  ` ...`<br> ` [0 0 0 0 0 0 0 0 0 1 0]]`     |\n",
        "| `x_skipgram_1hot` | `[[0 0 0 0 0 0 0 0 0 0 1]`<br> ` [0 0 0 0 0 0 0 0 0 0 1]`<br> ` [0 0 0 0 0 0 0 0 0 1 0]`<br>  ` ...`<br> ` [0 0 0 0 0 0 0 0 0 1 0]]` |\n",
        "| `y_skipgram_1hot` | `[[0 0 0 0 0 0 0 0 0 1 0]`<br> ` [0 0 0 0 0 0 0 1 0 0 0]`<br> ` [0 0 0 0 0 0 0 0 0 0 1]`<br>  ` ...`<br> ` [0 0 0 0 1 0 0 0 0 0 0]]`     |\n",
        "\n",
        "</center>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_pMiHbDDakY8",
        "outputId": "5aa92c52-ee6a-4ddd-8bbc-c8c63005b1c7"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[ 5]\n",
            " [43]\n",
            " [46]\n",
            " [63]]\n",
            "(227, 105)\n",
            "[[0]]\n",
            "(227, 105)\n",
            "[[63]]\n",
            "(902, 105)\n",
            "[[43]]\n",
            "(902, 105)\n"
          ]
        }
      ],
      "source": [
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "#-# STEP 4:  Generate training data part 2  #-#\n",
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "\n",
        "def one_hot(arr, num_vocabs):\n",
        "  \"\"\"\n",
        "  Takes in a 1D array and generates its one-hot representation.\n",
        "  Args:\n",
        "    arr: a 1D numpy array\n",
        "    num_vocabs: the number of unique vocabs\n",
        "  Returns:\n",
        "    a 2D numpy array of the one-hot representation of arr\n",
        "  \"\"\"\n",
        "  ### TO DO ###\n",
        "  onehot_arr = np.zeros((arr.size, num_vocabs))\n",
        "  onehot_arr[np.arange(arr.size), arr] = 1\n",
        "  return onehot_arr\n",
        "\n",
        "def one_hot_multicategorical(arr, num_vocabs):\n",
        "  \"\"\"\n",
        "  Takes in a 2D array and generates its multicategorical one-hot representation.\n",
        "  Args:\n",
        "    arr: a 2D numpy array\n",
        "    num_vocabs: the number of unique vocabs\n",
        "  Returns:\n",
        "    a 2D numpy array of the multicategorical one-hot representation of arr\n",
        "  \"\"\"\n",
        "  ### TO DO ###\n",
        "  #onehot_arr = np.zeros((arr.size[0], num_vocabs))\n",
        "  num_samples = arr.shape[0]\n",
        "  max_categories = arr.shape[1]\n",
        "  arr_1hot = np.zeros((num_samples, max_categories, num_vocabs+1), dtype= int)\n",
        "  arr_1hot[np.repeat(np.arange(num_samples), max_categories), np.tile(np.arange(max_categories), num_samples), arr.flatten()] = 1\n",
        "  arr_1hot = np.sum(arr_1hot, axis=1)\n",
        "  return arr_1hot[:,:-1]\n",
        "\n",
        "\n",
        "\n",
        "if __name__ == '__main__':\n",
        "  x_cbow_1hot = one_hot_multicategorical(x_cbow, num_vocabs)\n",
        "  y_cbow_1hot = one_hot(y_cbow, num_vocabs)\n",
        "  x_skipgram_1hot = one_hot(x_skipgram, num_vocabs)\n",
        "  y_skipgram_1hot = one_hot(y_skipgram, num_vocabs)\n",
        "  print(np.argwhere(x_cbow_1hot[10]))\n",
        "  print(x_cbow_1hot.shape)\n",
        "  print(np.argwhere(y_cbow_1hot[20]))\n",
        "  print(y_cbow_1hot.shape)\n",
        "  print(np.argwhere(x_skipgram_1hot[30]))\n",
        "  print(x_skipgram_1hot.shape)\n",
        "  print(np.argwhere(y_skipgram_1hot[40]))\n",
        "  print(y_skipgram_1hot.shape)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6ST9v8mp73MP"
      },
      "source": [
        "# **Section 3** Compiling and training your own Word2Vec model (2 tasks to complete)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EsLNyshbKEu4"
      },
      "source": [
        "After generating the dataset, it is time to build and train the model!  Using Keras, and create the following **two** variables,\n",
        "\n",
        "<center>\n",
        "\n",
        "| variable name    | datatype           | description                                                                                                                                                                                                                                                                                                                                                                                                                                                                       |\n",
        "|------------------|--------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|\n",
        "| `model_cbow`     | `keras.Sequential` | A Keras multilayer perceptron model with the following architecture:<br>1st layer: fully connected with `dims_to_learn` units, no activation function and no bias.  Specify the correct input shape.<br>2nd layer: fully connected with softmax activation function and no bias.  Specify the correct number of units.<br>|\n",
        "| `model_skipgram` | `keras.Sequential` | same as above   |\n",
        "\n",
        "</center>\n",
        "\n",
        "Also, compile both `model_cbow` and `model_skipgram` with the `Adam` gradient descent optimizer and the `categorical crossentropy` loss function."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5Z5_nS6DSvt8",
        "outputId": "7908e9da-cf77-4cba-afe4-3d18e8bc3258"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " dense (Dense)               (None, 50)                5250      \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 105)               5250      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 10500 (41.02 KB)\n",
            "Trainable params: 10500 (41.02 KB)\n",
            "Non-trainable params: 0 (0.00 Byte)\n",
            "_________________________________________________________________\n",
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " dense_2 (Dense)             (None, 50)                5250      \n",
            "                                                                 \n",
            " dense_3 (Dense)             (None, 105)               5250      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 10500 (41.02 KB)\n",
            "Trainable params: 10500 (41.02 KB)\n",
            "Non-trainable params: 0 (0.00 Byte)\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "#-# STEP 5:  Compile model  #-#\n",
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "\n",
        "import os\n",
        "os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'\n",
        "from keras import Sequential\n",
        "from keras.layers import Dense\n",
        "\n",
        "def create_cbow_model(num_vocabs, dims_to_learn=50):\n",
        "  ### TO DO ###\n",
        "  #model = Sequential(Dense(dims_to_learn, activation=None, use_bias=False, input_shape = (num_vocabs, ), name=\"layer1\"), Dense(num_units, activation=\"softmax\", use_bias=False, name=\"layer2\"))\n",
        "  model = Sequential()\n",
        "  model.add(Dense(units = dims_to_learn, activation = None, use_bias = False, input_shape = (num_vocabs, )))\n",
        "  model.add(Dense(units = num_vocabs, activation = 'softmax', use_bias = False ))\n",
        "\n",
        "  model.compile(optimizer='adam', loss='categorical_crossentropy',metrics = ['accuracy'])\n",
        "  return model\n",
        "def create_skipgram_model(num_vocabs, dims_to_learn=50):\n",
        "  ### TO DO ###\n",
        "  '''\n",
        "  model = Sequential()\n",
        "  model.add(Dense(units = dims_to_learn, activation = None, use_bias = False, input_shape = (num_vocabs, )))\n",
        "  model.add(Dense(units = num_vocabs, activation = 'softmax', use_bias = False ))\n",
        "  model.compile(optimizer='adam', loss='categorical_crossentropy')\n",
        "  '''\n",
        "  model = Sequential([\n",
        "        Dense(dims_to_learn, activation=None, use_bias=False, input_shape=(num_vocabs,)),\n",
        "        Dense(num_vocabs, activation='softmax', use_bias=False)\n",
        "    ])\n",
        "  model.compile(optimizer='adam', loss='categorical_crossentropy', metrics = ['accuracy'])\n",
        "  return model\n",
        "\n",
        "if __name__ == \"__main__\":\n",
        "  model_cbow = create_cbow_model(num_vocabs)\n",
        "  model_cbow.summary()\n",
        "  model_skipgram = create_skipgram_model(num_vocabs)\n",
        "  model_skipgram.summary()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9jhdXz6wraxu"
      },
      "source": [
        "Train both `model_cbow` and `model_skipgram` with their corresponding training data for 500 epochs."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "P9fqxJSwLT8V",
        "outputId": "3a6fdef5-bf92-4b84-b82f-0e401fa8b924"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/500\n",
            "8/8 [==============================] - 0s 12ms/step - loss: 0.0376 - accuracy: 0.9824\n",
            "Epoch 2/500\n",
            "8/8 [==============================] - 0s 12ms/step - loss: 0.0374 - accuracy: 0.9780\n",
            "Epoch 3/500\n",
            "8/8 [==============================] - 0s 11ms/step - loss: 0.0375 - accuracy: 0.9736\n",
            "Epoch 4/500\n",
            "8/8 [==============================] - 0s 12ms/step - loss: 0.0377 - accuracy: 0.9736\n",
            "Epoch 5/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0382 - accuracy: 0.9824\n",
            "Epoch 6/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0380 - accuracy: 0.9824\n",
            "Epoch 7/500\n",
            "8/8 [==============================] - 0s 10ms/step - loss: 0.0380 - accuracy: 0.9780\n",
            "Epoch 8/500\n",
            "8/8 [==============================] - 0s 13ms/step - loss: 0.0380 - accuracy: 0.9780\n",
            "Epoch 9/500\n",
            "8/8 [==============================] - 0s 10ms/step - loss: 0.0375 - accuracy: 0.9824\n",
            "Epoch 10/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0375 - accuracy: 0.9780\n",
            "Epoch 11/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0371 - accuracy: 0.9824\n",
            "Epoch 12/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0372 - accuracy: 0.9780\n",
            "Epoch 13/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0368 - accuracy: 0.9824\n",
            "Epoch 14/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0369 - accuracy: 0.9736\n",
            "Epoch 15/500\n",
            "8/8 [==============================] - 0s 9ms/step - loss: 0.0367 - accuracy: 0.9780\n",
            "Epoch 16/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0365 - accuracy: 0.9824\n",
            "Epoch 17/500\n",
            "8/8 [==============================] - 0s 9ms/step - loss: 0.0366 - accuracy: 0.9780\n",
            "Epoch 18/500\n",
            "8/8 [==============================] - 0s 11ms/step - loss: 0.0363 - accuracy: 0.9824\n",
            "Epoch 19/500\n",
            "8/8 [==============================] - 0s 9ms/step - loss: 0.0366 - accuracy: 0.9824\n",
            "Epoch 20/500\n",
            "8/8 [==============================] - 0s 9ms/step - loss: 0.0378 - accuracy: 0.9824\n",
            "Epoch 21/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0377 - accuracy: 0.9780\n",
            "Epoch 22/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0376 - accuracy: 0.9824\n",
            "Epoch 23/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0372 - accuracy: 0.9780\n",
            "Epoch 24/500\n",
            "8/8 [==============================] - 0s 10ms/step - loss: 0.0370 - accuracy: 0.9780\n",
            "Epoch 25/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0365 - accuracy: 0.9824\n",
            "Epoch 26/500\n",
            "8/8 [==============================] - 0s 8ms/step - loss: 0.0362 - accuracy: 0.9824\n",
            "Epoch 27/500\n",
            "8/8 [==============================] - 0s 9ms/step - loss: 0.0362 - accuracy: 0.9736\n",
            "Epoch 28/500\n",
            "8/8 [==============================] - 0s 13ms/step - loss: 0.0358 - accuracy: 0.9780\n",
            "Epoch 29/500\n",
            "8/8 [==============================] - 0s 9ms/step - loss: 0.0359 - accuracy: 0.9780\n",
            "Epoch 30/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0355 - accuracy: 0.9824\n",
            "Epoch 31/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0355 - accuracy: 0.9780\n",
            "Epoch 32/500\n",
            "8/8 [==============================] - 0s 9ms/step - loss: 0.0357 - accuracy: 0.9780\n",
            "Epoch 33/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0354 - accuracy: 0.9780\n",
            "Epoch 34/500\n",
            "8/8 [==============================] - 0s 10ms/step - loss: 0.0355 - accuracy: 0.9736\n",
            "Epoch 35/500\n",
            "8/8 [==============================] - 0s 9ms/step - loss: 0.0355 - accuracy: 0.9736\n",
            "Epoch 36/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0353 - accuracy: 0.9780\n",
            "Epoch 37/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0357 - accuracy: 0.9824\n",
            "Epoch 38/500\n",
            "8/8 [==============================] - 0s 8ms/step - loss: 0.0372 - accuracy: 0.9824\n",
            "Epoch 39/500\n",
            "8/8 [==============================] - 0s 12ms/step - loss: 0.0391 - accuracy: 0.9824\n",
            "Epoch 40/500\n",
            "8/8 [==============================] - 0s 16ms/step - loss: 0.0402 - accuracy: 0.9824\n",
            "Epoch 41/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0397 - accuracy: 0.9824\n",
            "Epoch 42/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0387 - accuracy: 0.9824\n",
            "Epoch 43/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0378 - accuracy: 0.9824\n",
            "Epoch 44/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0371 - accuracy: 0.9736\n",
            "Epoch 45/500\n",
            "8/8 [==============================] - 0s 9ms/step - loss: 0.0361 - accuracy: 0.9824\n",
            "Epoch 46/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0356 - accuracy: 0.9780\n",
            "Epoch 47/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0350 - accuracy: 0.9824\n",
            "Epoch 48/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0350 - accuracy: 0.9824\n",
            "Epoch 49/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0351 - accuracy: 0.9780\n",
            "Epoch 50/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0346 - accuracy: 0.9824\n",
            "Epoch 51/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0359 - accuracy: 0.9824\n",
            "Epoch 52/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0370 - accuracy: 0.9824\n",
            "Epoch 53/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0379 - accuracy: 0.9824\n",
            "Epoch 54/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0387 - accuracy: 0.9780\n",
            "Epoch 55/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0384 - accuracy: 0.9780\n",
            "Epoch 56/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0378 - accuracy: 0.9824\n",
            "Epoch 57/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0387 - accuracy: 0.9824\n",
            "Epoch 58/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0382 - accuracy: 0.9780\n",
            "Epoch 59/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0376 - accuracy: 0.9780\n",
            "Epoch 60/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0369 - accuracy: 0.9824\n",
            "Epoch 61/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0357 - accuracy: 0.9824\n",
            "Epoch 62/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0352 - accuracy: 0.9824\n",
            "Epoch 63/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0347 - accuracy: 0.9780\n",
            "Epoch 64/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0345 - accuracy: 0.9780\n",
            "Epoch 65/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0342 - accuracy: 0.9824\n",
            "Epoch 66/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0340 - accuracy: 0.9824\n",
            "Epoch 67/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0339 - accuracy: 0.9824\n",
            "Epoch 68/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0337 - accuracy: 0.9824\n",
            "Epoch 69/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0338 - accuracy: 0.9780\n",
            "Epoch 70/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0336 - accuracy: 0.9780\n",
            "Epoch 71/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0333 - accuracy: 0.9824\n",
            "Epoch 72/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0335 - accuracy: 0.9824\n",
            "Epoch 73/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0333 - accuracy: 0.9824\n",
            "Epoch 74/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0334 - accuracy: 0.9824\n",
            "Epoch 75/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0333 - accuracy: 0.9824\n",
            "Epoch 76/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0333 - accuracy: 0.9824\n",
            "Epoch 77/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0341 - accuracy: 0.9824\n",
            "Epoch 78/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0350 - accuracy: 0.9824\n",
            "Epoch 79/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0351 - accuracy: 0.9824\n",
            "Epoch 80/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0350 - accuracy: 0.9824\n",
            "Epoch 81/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0342 - accuracy: 0.9824\n",
            "Epoch 82/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0344 - accuracy: 0.9780\n",
            "Epoch 83/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0339 - accuracy: 0.9780\n",
            "Epoch 84/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0335 - accuracy: 0.9824\n",
            "Epoch 85/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0331 - accuracy: 0.9780\n",
            "Epoch 86/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0331 - accuracy: 0.9736\n",
            "Epoch 87/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0330 - accuracy: 0.9736\n",
            "Epoch 88/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0330 - accuracy: 0.9824\n",
            "Epoch 89/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0326 - accuracy: 0.9824\n",
            "Epoch 90/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0329 - accuracy: 0.9780\n",
            "Epoch 91/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0327 - accuracy: 0.9780\n",
            "Epoch 92/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0327 - accuracy: 0.9780\n",
            "Epoch 93/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0335 - accuracy: 0.9824\n",
            "Epoch 94/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0343 - accuracy: 0.9780\n",
            "Epoch 95/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0333 - accuracy: 0.9736\n",
            "Epoch 96/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0354 - accuracy: 0.9736\n",
            "Epoch 97/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0355 - accuracy: 0.9736\n",
            "Epoch 98/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0351 - accuracy: 0.9780\n",
            "Epoch 99/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0344 - accuracy: 0.9780\n",
            "Epoch 100/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0339 - accuracy: 0.9824\n",
            "Epoch 101/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0333 - accuracy: 0.9736\n",
            "Epoch 102/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0327 - accuracy: 0.9824\n",
            "Epoch 103/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0329 - accuracy: 0.9736\n",
            "Epoch 104/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0325 - accuracy: 0.9780\n",
            "Epoch 105/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0324 - accuracy: 0.9780\n",
            "Epoch 106/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0312 - accuracy: 0.9780\n",
            "Epoch 107/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0341 - accuracy: 0.9780\n",
            "Epoch 108/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0340 - accuracy: 0.9780\n",
            "Epoch 109/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0337 - accuracy: 0.9824\n",
            "Epoch 110/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0333 - accuracy: 0.9824\n",
            "Epoch 111/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0331 - accuracy: 0.9780\n",
            "Epoch 112/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0328 - accuracy: 0.9824\n",
            "Epoch 113/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0322 - accuracy: 0.9824\n",
            "Epoch 114/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0323 - accuracy: 0.9736\n",
            "Epoch 115/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0319 - accuracy: 0.9736\n",
            "Epoch 116/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0318 - accuracy: 0.9824\n",
            "Epoch 117/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0317 - accuracy: 0.9824\n",
            "Epoch 118/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0316 - accuracy: 0.9780\n",
            "Epoch 119/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0317 - accuracy: 0.9780\n",
            "Epoch 120/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0316 - accuracy: 0.9780\n",
            "Epoch 121/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0327 - accuracy: 0.9736\n",
            "Epoch 122/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0328 - accuracy: 0.9780\n",
            "Epoch 123/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0332 - accuracy: 0.9736\n",
            "Epoch 124/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0330 - accuracy: 0.9824\n",
            "Epoch 125/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0329 - accuracy: 0.9824\n",
            "Epoch 126/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0335 - accuracy: 0.9824\n",
            "Epoch 127/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0348 - accuracy: 0.9824\n",
            "Epoch 128/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0351 - accuracy: 0.9824\n",
            "Epoch 129/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0342 - accuracy: 0.9824\n",
            "Epoch 130/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0335 - accuracy: 0.9824\n",
            "Epoch 131/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0333 - accuracy: 0.9780\n",
            "Epoch 132/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0326 - accuracy: 0.9824\n",
            "Epoch 133/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0323 - accuracy: 0.9824\n",
            "Epoch 134/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0316 - accuracy: 0.9824\n",
            "Epoch 135/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0314 - accuracy: 0.9824\n",
            "Epoch 136/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0314 - accuracy: 0.9824\n",
            "Epoch 137/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0313 - accuracy: 0.9824\n",
            "Epoch 138/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0313 - accuracy: 0.9824\n",
            "Epoch 139/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0311 - accuracy: 0.9824\n",
            "Epoch 140/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0300 - accuracy: 0.9868\n",
            "Epoch 141/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0319 - accuracy: 0.9824\n",
            "Epoch 142/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0328 - accuracy: 0.9780\n",
            "Epoch 143/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0331 - accuracy: 0.9780\n",
            "Epoch 144/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0333 - accuracy: 0.9824\n",
            "Epoch 145/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0334 - accuracy: 0.9824\n",
            "Epoch 146/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0332 - accuracy: 0.9780\n",
            "Epoch 147/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0326 - accuracy: 0.9824\n",
            "Epoch 148/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0322 - accuracy: 0.9780\n",
            "Epoch 149/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0315 - accuracy: 0.9824\n",
            "Epoch 150/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0312 - accuracy: 0.9824\n",
            "Epoch 151/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0311 - accuracy: 0.9824\n",
            "Epoch 152/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0308 - accuracy: 0.9780\n",
            "Epoch 153/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0308 - accuracy: 0.9736\n",
            "Epoch 154/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0306 - accuracy: 0.9824\n",
            "Epoch 155/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0306 - accuracy: 0.9780\n",
            "Epoch 156/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0306 - accuracy: 0.9736\n",
            "Epoch 157/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0306 - accuracy: 0.9780\n",
            "Epoch 158/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0303 - accuracy: 0.9824\n",
            "Epoch 159/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0304 - accuracy: 0.9736\n",
            "Epoch 160/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0304 - accuracy: 0.9780\n",
            "Epoch 161/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0306 - accuracy: 0.9824\n",
            "Epoch 162/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0311 - accuracy: 0.9780\n",
            "Epoch 163/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0313 - accuracy: 0.9824\n",
            "Epoch 164/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0314 - accuracy: 0.9780\n",
            "Epoch 165/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0313 - accuracy: 0.9824\n",
            "Epoch 166/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0312 - accuracy: 0.9780\n",
            "Epoch 167/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0309 - accuracy: 0.9780\n",
            "Epoch 168/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0307 - accuracy: 0.9824\n",
            "Epoch 169/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0317 - accuracy: 0.9780\n",
            "Epoch 170/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0323 - accuracy: 0.9824\n",
            "Epoch 171/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0324 - accuracy: 0.9780\n",
            "Epoch 172/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0321 - accuracy: 0.9824\n",
            "Epoch 173/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0314 - accuracy: 0.9824\n",
            "Epoch 174/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0311 - accuracy: 0.9824\n",
            "Epoch 175/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0306 - accuracy: 0.9780\n",
            "Epoch 176/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0309 - accuracy: 0.9780\n",
            "Epoch 177/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0304 - accuracy: 0.9780\n",
            "Epoch 178/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0316 - accuracy: 0.9824\n",
            "Epoch 179/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0314 - accuracy: 0.9824\n",
            "Epoch 180/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0317 - accuracy: 0.9824\n",
            "Epoch 181/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0315 - accuracy: 0.9780\n",
            "Epoch 182/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0310 - accuracy: 0.9824\n",
            "Epoch 183/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0307 - accuracy: 0.9780\n",
            "Epoch 184/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0302 - accuracy: 0.9824\n",
            "Epoch 185/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0301 - accuracy: 0.9824\n",
            "Epoch 186/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0304 - accuracy: 0.9824\n",
            "Epoch 187/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0298 - accuracy: 0.9780\n",
            "Epoch 188/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0297 - accuracy: 0.9780\n",
            "Epoch 189/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0296 - accuracy: 0.9824\n",
            "Epoch 190/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0296 - accuracy: 0.9824\n",
            "Epoch 191/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0299 - accuracy: 0.9780\n",
            "Epoch 192/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0298 - accuracy: 0.9780\n",
            "Epoch 193/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0296 - accuracy: 0.9780\n",
            "Epoch 194/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0297 - accuracy: 0.9780\n",
            "Epoch 195/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0295 - accuracy: 0.9780\n",
            "Epoch 196/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0294 - accuracy: 0.9780\n",
            "Epoch 197/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0295 - accuracy: 0.9824\n",
            "Epoch 198/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0294 - accuracy: 0.9780\n",
            "Epoch 199/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0294 - accuracy: 0.9780\n",
            "Epoch 200/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0294 - accuracy: 0.9780\n",
            "Epoch 201/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0295 - accuracy: 0.9780\n",
            "Epoch 202/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0294 - accuracy: 0.9824\n",
            "Epoch 203/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0301 - accuracy: 0.9824\n",
            "Epoch 204/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0312 - accuracy: 0.9824\n",
            "Epoch 205/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0316 - accuracy: 0.9780\n",
            "Epoch 206/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0314 - accuracy: 0.9824\n",
            "Epoch 207/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0307 - accuracy: 0.9824\n",
            "Epoch 208/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0304 - accuracy: 0.9824\n",
            "Epoch 209/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0298 - accuracy: 0.9824\n",
            "Epoch 210/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0310 - accuracy: 0.9736\n",
            "Epoch 211/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0316 - accuracy: 0.9780\n",
            "Epoch 212/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0316 - accuracy: 0.9780\n",
            "Epoch 213/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0311 - accuracy: 0.9780\n",
            "Epoch 214/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0302 - accuracy: 0.9824\n",
            "Epoch 215/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0319 - accuracy: 0.9780\n",
            "Epoch 216/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0312 - accuracy: 0.9780\n",
            "Epoch 217/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0313 - accuracy: 0.9780\n",
            "Epoch 218/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0303 - accuracy: 0.9824\n",
            "Epoch 219/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0301 - accuracy: 0.9824\n",
            "Epoch 220/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0297 - accuracy: 0.9780\n",
            "Epoch 221/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0296 - accuracy: 0.9780\n",
            "Epoch 222/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0297 - accuracy: 0.9780\n",
            "Epoch 223/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0291 - accuracy: 0.9780\n",
            "Epoch 224/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0291 - accuracy: 0.9780\n",
            "Epoch 225/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0293 - accuracy: 0.9780\n",
            "Epoch 226/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0289 - accuracy: 0.9780\n",
            "Epoch 227/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0290 - accuracy: 0.9780\n",
            "Epoch 228/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0290 - accuracy: 0.9780\n",
            "Epoch 229/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0289 - accuracy: 0.9736\n",
            "Epoch 230/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 231/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0288 - accuracy: 0.9824\n",
            "Epoch 232/500\n",
            "8/8 [==============================] - 0s 9ms/step - loss: 0.0286 - accuracy: 0.9780\n",
            "Epoch 233/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0290 - accuracy: 0.9736\n",
            "Epoch 234/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 235/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0287 - accuracy: 0.9780\n",
            "Epoch 236/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0290 - accuracy: 0.9736\n",
            "Epoch 237/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0287 - accuracy: 0.9780\n",
            "Epoch 238/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0286 - accuracy: 0.9736\n",
            "Epoch 239/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0287 - accuracy: 0.9736\n",
            "Epoch 240/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0286 - accuracy: 0.9780\n",
            "Epoch 241/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0286 - accuracy: 0.9824\n",
            "Epoch 242/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0288 - accuracy: 0.9736\n",
            "Epoch 243/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0286 - accuracy: 0.9780\n",
            "Epoch 244/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0286 - accuracy: 0.9780\n",
            "Epoch 245/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0285 - accuracy: 0.9824\n",
            "Epoch 246/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0287 - accuracy: 0.9780\n",
            "Epoch 247/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0287 - accuracy: 0.9780\n",
            "Epoch 248/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0284 - accuracy: 0.9780\n",
            "Epoch 249/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0286 - accuracy: 0.9736\n",
            "Epoch 250/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0283 - accuracy: 0.9824\n",
            "Epoch 251/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0284 - accuracy: 0.9692\n",
            "Epoch 252/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0285 - accuracy: 0.9824\n",
            "Epoch 253/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0284 - accuracy: 0.9824\n",
            "Epoch 254/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0284 - accuracy: 0.9736\n",
            "Epoch 255/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0284 - accuracy: 0.9736\n",
            "Epoch 256/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0284 - accuracy: 0.9780\n",
            "Epoch 257/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0284 - accuracy: 0.9736\n",
            "Epoch 258/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0284 - accuracy: 0.9692\n",
            "Epoch 259/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0283 - accuracy: 0.9780\n",
            "Epoch 260/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0283 - accuracy: 0.9780\n",
            "Epoch 261/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0282 - accuracy: 0.9736\n",
            "Epoch 262/500\n",
            "8/8 [==============================] - 0s 8ms/step - loss: 0.0314 - accuracy: 0.9824\n",
            "Epoch 263/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0319 - accuracy: 0.9780\n",
            "Epoch 264/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0337 - accuracy: 0.9824\n",
            "Epoch 265/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0331 - accuracy: 0.9780\n",
            "Epoch 266/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0321 - accuracy: 0.9824\n",
            "Epoch 267/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0317 - accuracy: 0.9824\n",
            "Epoch 268/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0316 - accuracy: 0.9824\n",
            "Epoch 269/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0323 - accuracy: 0.9824\n",
            "Epoch 270/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0316 - accuracy: 0.9824\n",
            "Epoch 271/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0305 - accuracy: 0.9824\n",
            "Epoch 272/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0299 - accuracy: 0.9824\n",
            "Epoch 273/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0291 - accuracy: 0.9824\n",
            "Epoch 274/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0293 - accuracy: 0.9824\n",
            "Epoch 275/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0319 - accuracy: 0.9780\n",
            "Epoch 276/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0305 - accuracy: 0.9824\n",
            "Epoch 277/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0311 - accuracy: 0.9824\n",
            "Epoch 278/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0315 - accuracy: 0.9824\n",
            "Epoch 279/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0313 - accuracy: 0.9780\n",
            "Epoch 280/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0307 - accuracy: 0.9824\n",
            "Epoch 281/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0299 - accuracy: 0.9824\n",
            "Epoch 282/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0295 - accuracy: 0.9780\n",
            "Epoch 283/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0294 - accuracy: 0.9780\n",
            "Epoch 284/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0287 - accuracy: 0.9780\n",
            "Epoch 285/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0285 - accuracy: 0.9780\n",
            "Epoch 286/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0285 - accuracy: 0.9736\n",
            "Epoch 287/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0284 - accuracy: 0.9824\n",
            "Epoch 288/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0275 - accuracy: 0.9780\n",
            "Epoch 289/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0287 - accuracy: 0.9780\n",
            "Epoch 290/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0298 - accuracy: 0.9780\n",
            "Epoch 291/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0294 - accuracy: 0.9824\n",
            "Epoch 292/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0294 - accuracy: 0.9780\n",
            "Epoch 293/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0288 - accuracy: 0.9824\n",
            "Epoch 294/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0287 - accuracy: 0.9780\n",
            "Epoch 295/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0283 - accuracy: 0.9824\n",
            "Epoch 296/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0280 - accuracy: 0.9824\n",
            "Epoch 297/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0279 - accuracy: 0.9780\n",
            "Epoch 298/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0281 - accuracy: 0.9824\n",
            "Epoch 299/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0282 - accuracy: 0.9824\n",
            "Epoch 300/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0277 - accuracy: 0.9824\n",
            "Epoch 301/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0277 - accuracy: 0.9824\n",
            "Epoch 302/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0276 - accuracy: 0.9780\n",
            "Epoch 303/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0276 - accuracy: 0.9780\n",
            "Epoch 304/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0276 - accuracy: 0.9824\n",
            "Epoch 305/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0277 - accuracy: 0.9780\n",
            "Epoch 306/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0277 - accuracy: 0.9824\n",
            "Epoch 307/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0284 - accuracy: 0.9780\n",
            "Epoch 308/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0288 - accuracy: 0.9780\n",
            "Epoch 309/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0304 - accuracy: 0.9824\n",
            "Epoch 310/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0306 - accuracy: 0.9824\n",
            "Epoch 311/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0300 - accuracy: 0.9824\n",
            "Epoch 312/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0296 - accuracy: 0.9824\n",
            "Epoch 313/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0292 - accuracy: 0.9736\n",
            "Epoch 314/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 315/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0282 - accuracy: 0.9824\n",
            "Epoch 316/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0280 - accuracy: 0.9780\n",
            "Epoch 317/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0279 - accuracy: 0.9824\n",
            "Epoch 318/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0277 - accuracy: 0.9824\n",
            "Epoch 319/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0278 - accuracy: 0.9780\n",
            "Epoch 320/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0271 - accuracy: 0.9780\n",
            "Epoch 321/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0295 - accuracy: 0.9780\n",
            "Epoch 322/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0297 - accuracy: 0.9824\n",
            "Epoch 323/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0296 - accuracy: 0.9780\n",
            "Epoch 324/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0293 - accuracy: 0.9780\n",
            "Epoch 325/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0288 - accuracy: 0.9824\n",
            "Epoch 326/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0280 - accuracy: 0.9824\n",
            "Epoch 327/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0280 - accuracy: 0.9780\n",
            "Epoch 328/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0279 - accuracy: 0.9824\n",
            "Epoch 329/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0275 - accuracy: 0.9780\n",
            "Epoch 330/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0284 - accuracy: 0.9824\n",
            "Epoch 331/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0297 - accuracy: 0.9824\n",
            "Epoch 332/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0306 - accuracy: 0.9780\n",
            "Epoch 333/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0284 - accuracy: 0.9824\n",
            "Epoch 334/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0284 - accuracy: 0.9824\n",
            "Epoch 335/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0286 - accuracy: 0.9780\n",
            "Epoch 336/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0280 - accuracy: 0.9824\n",
            "Epoch 337/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0280 - accuracy: 0.9824\n",
            "Epoch 338/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0278 - accuracy: 0.9780\n",
            "Epoch 339/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0275 - accuracy: 0.9824\n",
            "Epoch 340/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0275 - accuracy: 0.9736\n",
            "Epoch 341/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0278 - accuracy: 0.9780\n",
            "Epoch 342/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0273 - accuracy: 0.9736\n",
            "Epoch 343/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0283 - accuracy: 0.9824\n",
            "Epoch 344/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 345/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 346/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0285 - accuracy: 0.9736\n",
            "Epoch 347/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0280 - accuracy: 0.9824\n",
            "Epoch 348/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0280 - accuracy: 0.9780\n",
            "Epoch 349/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0277 - accuracy: 0.9780\n",
            "Epoch 350/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0275 - accuracy: 0.9824\n",
            "Epoch 351/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0272 - accuracy: 0.9824\n",
            "Epoch 352/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0283 - accuracy: 0.9780\n",
            "Epoch 353/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0290 - accuracy: 0.9824\n",
            "Epoch 354/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0289 - accuracy: 0.9824\n",
            "Epoch 355/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0288 - accuracy: 0.9824\n",
            "Epoch 356/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0279 - accuracy: 0.9824\n",
            "Epoch 357/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0282 - accuracy: 0.9824\n",
            "Epoch 358/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 359/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0285 - accuracy: 0.9824\n",
            "Epoch 360/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0283 - accuracy: 0.9824\n",
            "Epoch 361/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0281 - accuracy: 0.9780\n",
            "Epoch 362/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0276 - accuracy: 0.9824\n",
            "Epoch 363/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0277 - accuracy: 0.9824\n",
            "Epoch 364/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0272 - accuracy: 0.9780\n",
            "Epoch 365/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0285 - accuracy: 0.9824\n",
            "Epoch 366/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0292 - accuracy: 0.9780\n",
            "Epoch 367/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0291 - accuracy: 0.9736\n",
            "Epoch 368/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0287 - accuracy: 0.9780\n",
            "Epoch 369/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0283 - accuracy: 0.9824\n",
            "Epoch 370/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0279 - accuracy: 0.9824\n",
            "Epoch 371/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0276 - accuracy: 0.9824\n",
            "Epoch 372/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0271 - accuracy: 0.9824\n",
            "Epoch 373/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0271 - accuracy: 0.9824\n",
            "Epoch 374/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0271 - accuracy: 0.9780\n",
            "Epoch 375/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0269 - accuracy: 0.9736\n",
            "Epoch 376/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0270 - accuracy: 0.9780\n",
            "Epoch 377/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0268 - accuracy: 0.9780\n",
            "Epoch 378/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0269 - accuracy: 0.9824\n",
            "Epoch 379/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0268 - accuracy: 0.9824\n",
            "Epoch 380/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0268 - accuracy: 0.9824\n",
            "Epoch 381/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0268 - accuracy: 0.9780\n",
            "Epoch 382/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0271 - accuracy: 0.9780\n",
            "Epoch 383/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0274 - accuracy: 0.9780\n",
            "Epoch 384/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0279 - accuracy: 0.9824\n",
            "Epoch 385/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0279 - accuracy: 0.9780\n",
            "Epoch 386/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0278 - accuracy: 0.9824\n",
            "Epoch 387/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0283 - accuracy: 0.9780\n",
            "Epoch 388/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0287 - accuracy: 0.9780\n",
            "Epoch 389/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0292 - accuracy: 0.9780\n",
            "Epoch 390/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0299 - accuracy: 0.9824\n",
            "Epoch 391/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0295 - accuracy: 0.9824\n",
            "Epoch 392/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0302 - accuracy: 0.9824\n",
            "Epoch 393/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0294 - accuracy: 0.9824\n",
            "Epoch 394/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0284 - accuracy: 0.9824\n",
            "Epoch 395/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0282 - accuracy: 0.9824\n",
            "Epoch 396/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0288 - accuracy: 0.9780\n",
            "Epoch 397/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0286 - accuracy: 0.9780\n",
            "Epoch 398/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0284 - accuracy: 0.9780\n",
            "Epoch 399/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0283 - accuracy: 0.9780\n",
            "Epoch 400/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0277 - accuracy: 0.9824\n",
            "Epoch 401/500\n",
            "8/8 [==============================] - 0s 7ms/step - loss: 0.0274 - accuracy: 0.9780\n",
            "Epoch 402/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0272 - accuracy: 0.9824\n",
            "Epoch 403/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0272 - accuracy: 0.9824\n",
            "Epoch 404/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0268 - accuracy: 0.9824\n",
            "Epoch 405/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0270 - accuracy: 0.9736\n",
            "Epoch 406/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0259 - accuracy: 0.9736\n",
            "Epoch 407/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0278 - accuracy: 0.9780\n",
            "Epoch 408/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 409/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 410/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0294 - accuracy: 0.9780\n",
            "Epoch 411/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 412/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0284 - accuracy: 0.9780\n",
            "Epoch 413/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0282 - accuracy: 0.9780\n",
            "Epoch 414/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0278 - accuracy: 0.9736\n",
            "Epoch 415/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0270 - accuracy: 0.9824\n",
            "Epoch 416/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0273 - accuracy: 0.9824\n",
            "Epoch 417/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0270 - accuracy: 0.9824\n",
            "Epoch 418/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0268 - accuracy: 0.9780\n",
            "Epoch 419/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0286 - accuracy: 0.9780\n",
            "Epoch 420/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0281 - accuracy: 0.9736\n",
            "Epoch 421/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0290 - accuracy: 0.9824\n",
            "Epoch 422/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 423/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0283 - accuracy: 0.9780\n",
            "Epoch 424/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0282 - accuracy: 0.9780\n",
            "Epoch 425/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0274 - accuracy: 0.9824\n",
            "Epoch 426/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0270 - accuracy: 0.9824\n",
            "Epoch 427/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0270 - accuracy: 0.9780\n",
            "Epoch 428/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0267 - accuracy: 0.9780\n",
            "Epoch 429/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0252 - accuracy: 0.9868\n",
            "Epoch 430/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0270 - accuracy: 0.9824\n",
            "Epoch 431/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0283 - accuracy: 0.9824\n",
            "Epoch 432/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0286 - accuracy: 0.9780\n",
            "Epoch 433/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0284 - accuracy: 0.9780\n",
            "Epoch 434/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0279 - accuracy: 0.9780\n",
            "Epoch 435/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0284 - accuracy: 0.9736\n",
            "Epoch 436/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0283 - accuracy: 0.9780\n",
            "Epoch 437/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0286 - accuracy: 0.9824\n",
            "Epoch 438/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0284 - accuracy: 0.9780\n",
            "Epoch 439/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0281 - accuracy: 0.9736\n",
            "Epoch 440/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0291 - accuracy: 0.9824\n",
            "Epoch 441/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0293 - accuracy: 0.9824\n",
            "Epoch 442/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0290 - accuracy: 0.9780\n",
            "Epoch 443/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 444/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0284 - accuracy: 0.9824\n",
            "Epoch 445/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0275 - accuracy: 0.9824\n",
            "Epoch 446/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0272 - accuracy: 0.9824\n",
            "Epoch 447/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0271 - accuracy: 0.9824\n",
            "Epoch 448/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0267 - accuracy: 0.9780\n",
            "Epoch 449/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0267 - accuracy: 0.9780\n",
            "Epoch 450/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0265 - accuracy: 0.9824\n",
            "Epoch 451/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0265 - accuracy: 0.9824\n",
            "Epoch 452/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0263 - accuracy: 0.9824\n",
            "Epoch 453/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0263 - accuracy: 0.9780\n",
            "Epoch 454/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0262 - accuracy: 0.9824\n",
            "Epoch 455/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0262 - accuracy: 0.9780\n",
            "Epoch 456/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0264 - accuracy: 0.9780\n",
            "Epoch 457/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0264 - accuracy: 0.9780\n",
            "Epoch 458/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0262 - accuracy: 0.9780\n",
            "Epoch 459/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0264 - accuracy: 0.9736\n",
            "Epoch 460/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0263 - accuracy: 0.9780\n",
            "Epoch 461/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0264 - accuracy: 0.9780\n",
            "Epoch 462/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0271 - accuracy: 0.9780\n",
            "Epoch 463/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0278 - accuracy: 0.9780\n",
            "Epoch 464/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0277 - accuracy: 0.9824\n",
            "Epoch 465/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0276 - accuracy: 0.9780\n",
            "Epoch 466/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0270 - accuracy: 0.9780\n",
            "Epoch 467/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0266 - accuracy: 0.9736\n",
            "Epoch 468/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0270 - accuracy: 0.9824\n",
            "Epoch 469/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0274 - accuracy: 0.9824\n",
            "Epoch 470/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0273 - accuracy: 0.9824\n",
            "Epoch 471/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0277 - accuracy: 0.9824\n",
            "Epoch 472/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 473/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0287 - accuracy: 0.9824\n",
            "Epoch 474/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0284 - accuracy: 0.9824\n",
            "Epoch 475/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0283 - accuracy: 0.9824\n",
            "Epoch 476/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0285 - accuracy: 0.9824\n",
            "Epoch 477/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0285 - accuracy: 0.9824\n",
            "Epoch 478/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0283 - accuracy: 0.9824\n",
            "Epoch 479/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0280 - accuracy: 0.9824\n",
            "Epoch 480/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0272 - accuracy: 0.9824\n",
            "Epoch 481/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0270 - accuracy: 0.9824\n",
            "Epoch 482/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0268 - accuracy: 0.9824\n",
            "Epoch 483/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0268 - accuracy: 0.9824\n",
            "Epoch 484/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0264 - accuracy: 0.9824\n",
            "Epoch 485/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0264 - accuracy: 0.9824\n",
            "Epoch 486/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0262 - accuracy: 0.9780\n",
            "Epoch 487/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0263 - accuracy: 0.9824\n",
            "Epoch 488/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0261 - accuracy: 0.9780\n",
            "Epoch 489/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0262 - accuracy: 0.9780\n",
            "Epoch 490/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0270 - accuracy: 0.9824\n",
            "Epoch 491/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0279 - accuracy: 0.9824\n",
            "Epoch 492/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0283 - accuracy: 0.9824\n",
            "Epoch 493/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0281 - accuracy: 0.9824\n",
            "Epoch 494/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0276 - accuracy: 0.9824\n",
            "Epoch 495/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0274 - accuracy: 0.9824\n",
            "Epoch 496/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0269 - accuracy: 0.9736\n",
            "Epoch 497/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0264 - accuracy: 0.9824\n",
            "Epoch 498/500\n",
            "8/8 [==============================] - 0s 4ms/step - loss: 0.0267 - accuracy: 0.9824\n",
            "Epoch 499/500\n",
            "8/8 [==============================] - 0s 5ms/step - loss: 0.0277 - accuracy: 0.9824\n",
            "Epoch 500/500\n",
            "8/8 [==============================] - 0s 6ms/step - loss: 0.0284 - accuracy: 0.9824\n",
            "Epoch 1/500\n",
            "29/29 [==============================] - 1s 5ms/step - loss: 4.6536 - accuracy: 0.0122\n",
            "Epoch 2/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 4.6267 - accuracy: 0.0200\n",
            "Epoch 3/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 4.6016 - accuracy: 0.0388\n",
            "Epoch 4/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 4.5747 - accuracy: 0.0676\n",
            "Epoch 5/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 4.5457 - accuracy: 0.0965\n",
            "Epoch 6/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 4.5132 - accuracy: 0.1131\n",
            "Epoch 7/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 4.4762 - accuracy: 0.1297\n",
            "Epoch 8/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 4.4345 - accuracy: 0.1375\n",
            "Epoch 9/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 4.3873 - accuracy: 0.1519\n",
            "Epoch 10/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 4.3342 - accuracy: 0.1574\n",
            "Epoch 11/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 4.2758 - accuracy: 0.1619\n",
            "Epoch 12/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 4.2140 - accuracy: 0.1641\n",
            "Epoch 13/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 4.1485 - accuracy: 0.1763\n",
            "Epoch 14/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 4.0825 - accuracy: 0.1752\n",
            "Epoch 15/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 4.0162 - accuracy: 0.1785\n",
            "Epoch 16/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.9523 - accuracy: 0.1774\n",
            "Epoch 17/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.8911 - accuracy: 0.1752\n",
            "Epoch 18/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.8329 - accuracy: 0.1818\n",
            "Epoch 19/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.7763 - accuracy: 0.1785\n",
            "Epoch 20/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.7209 - accuracy: 0.1818\n",
            "Epoch 21/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.6660 - accuracy: 0.1829\n",
            "Epoch 22/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.6138 - accuracy: 0.1818\n",
            "Epoch 23/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 3.5616 - accuracy: 0.1885\n",
            "Epoch 24/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.5107 - accuracy: 0.1885\n",
            "Epoch 25/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 3.4607 - accuracy: 0.1885\n",
            "Epoch 26/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.4117 - accuracy: 0.1896\n",
            "Epoch 27/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 3.3646 - accuracy: 0.1863\n",
            "Epoch 28/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.3174 - accuracy: 0.1896\n",
            "Epoch 29/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 3.2724 - accuracy: 0.1929\n",
            "Epoch 30/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 3.2277 - accuracy: 0.1929\n",
            "Epoch 31/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.1864 - accuracy: 0.1896\n",
            "Epoch 32/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 3.1455 - accuracy: 0.1896\n",
            "Epoch 33/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 3.1065 - accuracy: 0.1918\n",
            "Epoch 34/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 3.0699 - accuracy: 0.1929\n",
            "Epoch 35/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 3.0334 - accuracy: 0.1896\n",
            "Epoch 36/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.9989 - accuracy: 0.1918\n",
            "Epoch 37/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.9662 - accuracy: 0.1851\n",
            "Epoch 38/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.9353 - accuracy: 0.1874\n",
            "Epoch 39/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.9049 - accuracy: 0.1918\n",
            "Epoch 40/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.8764 - accuracy: 0.1940\n",
            "Epoch 41/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.8492 - accuracy: 0.1929\n",
            "Epoch 42/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.8233 - accuracy: 0.1940\n",
            "Epoch 43/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.7986 - accuracy: 0.1918\n",
            "Epoch 44/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.7749 - accuracy: 0.1929\n",
            "Epoch 45/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.7532 - accuracy: 0.1907\n",
            "Epoch 46/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.7320 - accuracy: 0.1951\n",
            "Epoch 47/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.7124 - accuracy: 0.1907\n",
            "Epoch 48/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.6930 - accuracy: 0.1907\n",
            "Epoch 49/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.6751 - accuracy: 0.1896\n",
            "Epoch 50/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.6581 - accuracy: 0.1885\n",
            "Epoch 51/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.6417 - accuracy: 0.1918\n",
            "Epoch 52/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.6269 - accuracy: 0.1918\n",
            "Epoch 53/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.6117 - accuracy: 0.1929\n",
            "Epoch 54/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.5983 - accuracy: 0.1863\n",
            "Epoch 55/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.5847 - accuracy: 0.1929\n",
            "Epoch 56/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.5727 - accuracy: 0.1840\n",
            "Epoch 57/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.5621 - accuracy: 0.1874\n",
            "Epoch 58/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.5497 - accuracy: 0.1807\n",
            "Epoch 59/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.5388 - accuracy: 0.1907\n",
            "Epoch 60/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.5286 - accuracy: 0.1918\n",
            "Epoch 61/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.5187 - accuracy: 0.1874\n",
            "Epoch 62/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.5106 - accuracy: 0.1896\n",
            "Epoch 63/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.5010 - accuracy: 0.1885\n",
            "Epoch 64/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.4925 - accuracy: 0.1863\n",
            "Epoch 65/500\n",
            "29/29 [==============================] - 0s 10ms/step - loss: 2.4834 - accuracy: 0.1796\n",
            "Epoch 66/500\n",
            "29/29 [==============================] - 0s 8ms/step - loss: 2.4756 - accuracy: 0.1829\n",
            "Epoch 67/500\n",
            "29/29 [==============================] - 0s 9ms/step - loss: 2.4681 - accuracy: 0.1874\n",
            "Epoch 68/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.4622 - accuracy: 0.1874\n",
            "Epoch 69/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.4553 - accuracy: 0.1840\n",
            "Epoch 70/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.4479 - accuracy: 0.1885\n",
            "Epoch 71/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.4413 - accuracy: 0.1896\n",
            "Epoch 72/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.4355 - accuracy: 0.1829\n",
            "Epoch 73/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.4302 - accuracy: 0.1863\n",
            "Epoch 74/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.4244 - accuracy: 0.1807\n",
            "Epoch 75/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.4181 - accuracy: 0.1874\n",
            "Epoch 76/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.4140 - accuracy: 0.1896\n",
            "Epoch 77/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.4095 - accuracy: 0.1785\n",
            "Epoch 78/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.4038 - accuracy: 0.1807\n",
            "Epoch 79/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3990 - accuracy: 0.1785\n",
            "Epoch 80/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3953 - accuracy: 0.1796\n",
            "Epoch 81/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3910 - accuracy: 0.1785\n",
            "Epoch 82/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3862 - accuracy: 0.1763\n",
            "Epoch 83/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3821 - accuracy: 0.1729\n",
            "Epoch 84/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3794 - accuracy: 0.1829\n",
            "Epoch 85/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3761 - accuracy: 0.1774\n",
            "Epoch 86/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3723 - accuracy: 0.1785\n",
            "Epoch 87/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3690 - accuracy: 0.1785\n",
            "Epoch 88/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3656 - accuracy: 0.1785\n",
            "Epoch 89/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3624 - accuracy: 0.1774\n",
            "Epoch 90/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3595 - accuracy: 0.1741\n",
            "Epoch 91/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3567 - accuracy: 0.1696\n",
            "Epoch 92/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3530 - accuracy: 0.1763\n",
            "Epoch 93/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3517 - accuracy: 0.1796\n",
            "Epoch 94/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3490 - accuracy: 0.1796\n",
            "Epoch 95/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3467 - accuracy: 0.1752\n",
            "Epoch 96/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3436 - accuracy: 0.1785\n",
            "Epoch 97/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.3415 - accuracy: 0.1763\n",
            "Epoch 98/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3391 - accuracy: 0.1752\n",
            "Epoch 99/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.3366 - accuracy: 0.1741\n",
            "Epoch 100/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.3351 - accuracy: 0.1752\n",
            "Epoch 101/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3327 - accuracy: 0.1707\n",
            "Epoch 102/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3305 - accuracy: 0.1652\n",
            "Epoch 103/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.3283 - accuracy: 0.1718\n",
            "Epoch 104/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3260 - accuracy: 0.1763\n",
            "Epoch 105/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3245 - accuracy: 0.1752\n",
            "Epoch 106/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3228 - accuracy: 0.1785\n",
            "Epoch 107/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3217 - accuracy: 0.1696\n",
            "Epoch 108/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3192 - accuracy: 0.1718\n",
            "Epoch 109/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3171 - accuracy: 0.1774\n",
            "Epoch 110/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3163 - accuracy: 0.1774\n",
            "Epoch 111/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3144 - accuracy: 0.1796\n",
            "Epoch 112/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3133 - accuracy: 0.1663\n",
            "Epoch 113/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3125 - accuracy: 0.1718\n",
            "Epoch 114/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3110 - accuracy: 0.1741\n",
            "Epoch 115/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3087 - accuracy: 0.1696\n",
            "Epoch 116/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3071 - accuracy: 0.1718\n",
            "Epoch 117/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3063 - accuracy: 0.1663\n",
            "Epoch 118/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3056 - accuracy: 0.1729\n",
            "Epoch 119/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.3036 - accuracy: 0.1785\n",
            "Epoch 120/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3027 - accuracy: 0.1718\n",
            "Epoch 121/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3009 - accuracy: 0.1718\n",
            "Epoch 122/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.3008 - accuracy: 0.1663\n",
            "Epoch 123/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2986 - accuracy: 0.1674\n",
            "Epoch 124/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2986 - accuracy: 0.1774\n",
            "Epoch 125/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2971 - accuracy: 0.1652\n",
            "Epoch 126/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2967 - accuracy: 0.1718\n",
            "Epoch 127/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2951 - accuracy: 0.1652\n",
            "Epoch 128/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2941 - accuracy: 0.1685\n",
            "Epoch 129/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2933 - accuracy: 0.1774\n",
            "Epoch 130/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2922 - accuracy: 0.1729\n",
            "Epoch 131/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2925 - accuracy: 0.1663\n",
            "Epoch 132/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2908 - accuracy: 0.1696\n",
            "Epoch 133/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2894 - accuracy: 0.1619\n",
            "Epoch 134/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2885 - accuracy: 0.1596\n",
            "Epoch 135/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2868 - accuracy: 0.1563\n",
            "Epoch 136/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2871 - accuracy: 0.1663\n",
            "Epoch 137/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2859 - accuracy: 0.1619\n",
            "Epoch 138/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2850 - accuracy: 0.1641\n",
            "Epoch 139/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2839 - accuracy: 0.1685\n",
            "Epoch 140/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2836 - accuracy: 0.1608\n",
            "Epoch 141/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2834 - accuracy: 0.1630\n",
            "Epoch 142/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2817 - accuracy: 0.1674\n",
            "Epoch 143/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2818 - accuracy: 0.1641\n",
            "Epoch 144/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2806 - accuracy: 0.1630\n",
            "Epoch 145/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2800 - accuracy: 0.1685\n",
            "Epoch 146/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2794 - accuracy: 0.1707\n",
            "Epoch 147/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2785 - accuracy: 0.1541\n",
            "Epoch 148/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2786 - accuracy: 0.1663\n",
            "Epoch 149/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2770 - accuracy: 0.1608\n",
            "Epoch 150/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2780 - accuracy: 0.1652\n",
            "Epoch 151/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2759 - accuracy: 0.1641\n",
            "Epoch 152/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2771 - accuracy: 0.1652\n",
            "Epoch 153/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2758 - accuracy: 0.1619\n",
            "Epoch 154/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2746 - accuracy: 0.1619\n",
            "Epoch 155/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2748 - accuracy: 0.1608\n",
            "Epoch 156/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2741 - accuracy: 0.1630\n",
            "Epoch 157/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2728 - accuracy: 0.1741\n",
            "Epoch 158/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2718 - accuracy: 0.1729\n",
            "Epoch 159/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2721 - accuracy: 0.1641\n",
            "Epoch 160/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2722 - accuracy: 0.1674\n",
            "Epoch 161/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2707 - accuracy: 0.1663\n",
            "Epoch 162/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2710 - accuracy: 0.1652\n",
            "Epoch 163/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2709 - accuracy: 0.1696\n",
            "Epoch 164/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2692 - accuracy: 0.1630\n",
            "Epoch 165/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2689 - accuracy: 0.1608\n",
            "Epoch 166/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2683 - accuracy: 0.1552\n",
            "Epoch 167/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2689 - accuracy: 0.1630\n",
            "Epoch 168/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2679 - accuracy: 0.1641\n",
            "Epoch 169/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2680 - accuracy: 0.1563\n",
            "Epoch 170/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2679 - accuracy: 0.1585\n",
            "Epoch 171/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2667 - accuracy: 0.1630\n",
            "Epoch 172/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2662 - accuracy: 0.1685\n",
            "Epoch 173/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2660 - accuracy: 0.1652\n",
            "Epoch 174/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2654 - accuracy: 0.1596\n",
            "Epoch 175/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2655 - accuracy: 0.1663\n",
            "Epoch 176/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2651 - accuracy: 0.1508\n",
            "Epoch 177/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2638 - accuracy: 0.1619\n",
            "Epoch 178/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2646 - accuracy: 0.1608\n",
            "Epoch 179/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2639 - accuracy: 0.1652\n",
            "Epoch 180/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2631 - accuracy: 0.1652\n",
            "Epoch 181/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2626 - accuracy: 0.1574\n",
            "Epoch 182/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2630 - accuracy: 0.1641\n",
            "Epoch 183/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2622 - accuracy: 0.1674\n",
            "Epoch 184/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2627 - accuracy: 0.1596\n",
            "Epoch 185/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2619 - accuracy: 0.1585\n",
            "Epoch 186/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2615 - accuracy: 0.1630\n",
            "Epoch 187/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2607 - accuracy: 0.1663\n",
            "Epoch 188/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2606 - accuracy: 0.1585\n",
            "Epoch 189/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2598 - accuracy: 0.1663\n",
            "Epoch 190/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2605 - accuracy: 0.1574\n",
            "Epoch 191/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2600 - accuracy: 0.1663\n",
            "Epoch 192/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2604 - accuracy: 0.1585\n",
            "Epoch 193/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2597 - accuracy: 0.1585\n",
            "Epoch 194/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2593 - accuracy: 0.1663\n",
            "Epoch 195/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2584 - accuracy: 0.1585\n",
            "Epoch 196/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2588 - accuracy: 0.1574\n",
            "Epoch 197/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2585 - accuracy: 0.1674\n",
            "Epoch 198/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2579 - accuracy: 0.1630\n",
            "Epoch 199/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2578 - accuracy: 0.1596\n",
            "Epoch 200/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2568 - accuracy: 0.1619\n",
            "Epoch 201/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2565 - accuracy: 0.1596\n",
            "Epoch 202/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2564 - accuracy: 0.1596\n",
            "Epoch 203/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2568 - accuracy: 0.1652\n",
            "Epoch 204/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2570 - accuracy: 0.1685\n",
            "Epoch 205/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2566 - accuracy: 0.1641\n",
            "Epoch 206/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2570 - accuracy: 0.1596\n",
            "Epoch 207/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2559 - accuracy: 0.1685\n",
            "Epoch 208/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2561 - accuracy: 0.1585\n",
            "Epoch 209/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2551 - accuracy: 0.1563\n",
            "Epoch 210/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2552 - accuracy: 0.1630\n",
            "Epoch 211/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2548 - accuracy: 0.1707\n",
            "Epoch 212/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2555 - accuracy: 0.1552\n",
            "Epoch 213/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2552 - accuracy: 0.1541\n",
            "Epoch 214/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2547 - accuracy: 0.1641\n",
            "Epoch 215/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2540 - accuracy: 0.1696\n",
            "Epoch 216/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2538 - accuracy: 0.1596\n",
            "Epoch 217/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2535 - accuracy: 0.1574\n",
            "Epoch 218/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2544 - accuracy: 0.1641\n",
            "Epoch 219/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2526 - accuracy: 0.1718\n",
            "Epoch 220/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2532 - accuracy: 0.1663\n",
            "Epoch 221/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2531 - accuracy: 0.1519\n",
            "Epoch 222/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2530 - accuracy: 0.1619\n",
            "Epoch 223/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2526 - accuracy: 0.1541\n",
            "Epoch 224/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2523 - accuracy: 0.1552\n",
            "Epoch 225/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2521 - accuracy: 0.1552\n",
            "Epoch 226/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2518 - accuracy: 0.1619\n",
            "Epoch 227/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2524 - accuracy: 0.1608\n",
            "Epoch 228/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2517 - accuracy: 0.1663\n",
            "Epoch 229/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2510 - accuracy: 0.1652\n",
            "Epoch 230/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2520 - accuracy: 0.1608\n",
            "Epoch 231/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2514 - accuracy: 0.1685\n",
            "Epoch 232/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2513 - accuracy: 0.1585\n",
            "Epoch 233/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2508 - accuracy: 0.1608\n",
            "Epoch 234/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2508 - accuracy: 0.1619\n",
            "Epoch 235/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2505 - accuracy: 0.1608\n",
            "Epoch 236/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2507 - accuracy: 0.1619\n",
            "Epoch 237/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2498 - accuracy: 0.1652\n",
            "Epoch 238/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2501 - accuracy: 0.1641\n",
            "Epoch 239/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2490 - accuracy: 0.1641\n",
            "Epoch 240/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2509 - accuracy: 0.1685\n",
            "Epoch 241/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2490 - accuracy: 0.1619\n",
            "Epoch 242/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2484 - accuracy: 0.1707\n",
            "Epoch 243/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2493 - accuracy: 0.1641\n",
            "Epoch 244/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2487 - accuracy: 0.1685\n",
            "Epoch 245/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2495 - accuracy: 0.1574\n",
            "Epoch 246/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2489 - accuracy: 0.1630\n",
            "Epoch 247/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2487 - accuracy: 0.1652\n",
            "Epoch 248/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2482 - accuracy: 0.1608\n",
            "Epoch 249/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2474 - accuracy: 0.1685\n",
            "Epoch 250/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2483 - accuracy: 0.1608\n",
            "Epoch 251/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2484 - accuracy: 0.1596\n",
            "Epoch 252/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2481 - accuracy: 0.1585\n",
            "Epoch 253/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2479 - accuracy: 0.1685\n",
            "Epoch 254/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2479 - accuracy: 0.1608\n",
            "Epoch 255/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2486 - accuracy: 0.1552\n",
            "Epoch 256/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2474 - accuracy: 0.1596\n",
            "Epoch 257/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2470 - accuracy: 0.1596\n",
            "Epoch 258/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2474 - accuracy: 0.1530\n",
            "Epoch 259/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2467 - accuracy: 0.1663\n",
            "Epoch 260/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2465 - accuracy: 0.1563\n",
            "Epoch 261/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2464 - accuracy: 0.1563\n",
            "Epoch 262/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2462 - accuracy: 0.1563\n",
            "Epoch 263/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2467 - accuracy: 0.1608\n",
            "Epoch 264/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2460 - accuracy: 0.1530\n",
            "Epoch 265/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2460 - accuracy: 0.1452\n",
            "Epoch 266/500\n",
            "29/29 [==============================] - 0s 3ms/step - loss: 2.2462 - accuracy: 0.1596\n",
            "Epoch 267/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2460 - accuracy: 0.1652\n",
            "Epoch 268/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2459 - accuracy: 0.1541\n",
            "Epoch 269/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2459 - accuracy: 0.1585\n",
            "Epoch 270/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2451 - accuracy: 0.1674\n",
            "Epoch 271/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2449 - accuracy: 0.1541\n",
            "Epoch 272/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2451 - accuracy: 0.1585\n",
            "Epoch 273/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2453 - accuracy: 0.1585\n",
            "Epoch 274/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2452 - accuracy: 0.1574\n",
            "Epoch 275/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2450 - accuracy: 0.1630\n",
            "Epoch 276/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2442 - accuracy: 0.1663\n",
            "Epoch 277/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2448 - accuracy: 0.1696\n",
            "Epoch 278/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2445 - accuracy: 0.1596\n",
            "Epoch 279/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2445 - accuracy: 0.1619\n",
            "Epoch 280/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2441 - accuracy: 0.1563\n",
            "Epoch 281/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2442 - accuracy: 0.1574\n",
            "Epoch 282/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2437 - accuracy: 0.1630\n",
            "Epoch 283/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2446 - accuracy: 0.1663\n",
            "Epoch 284/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2436 - accuracy: 0.1652\n",
            "Epoch 285/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2441 - accuracy: 0.1619\n",
            "Epoch 286/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2434 - accuracy: 0.1630\n",
            "Epoch 287/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2446 - accuracy: 0.1674\n",
            "Epoch 288/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2438 - accuracy: 0.1519\n",
            "Epoch 289/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2445 - accuracy: 0.1574\n",
            "Epoch 290/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2434 - accuracy: 0.1685\n",
            "Epoch 291/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2433 - accuracy: 0.1652\n",
            "Epoch 292/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2426 - accuracy: 0.1596\n",
            "Epoch 293/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2429 - accuracy: 0.1608\n",
            "Epoch 294/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2426 - accuracy: 0.1530\n",
            "Epoch 295/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2434 - accuracy: 0.1619\n",
            "Epoch 296/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2425 - accuracy: 0.1663\n",
            "Epoch 297/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2421 - accuracy: 0.1652\n",
            "Epoch 298/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2421 - accuracy: 0.1685\n",
            "Epoch 299/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2424 - accuracy: 0.1574\n",
            "Epoch 300/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2428 - accuracy: 0.1641\n",
            "Epoch 301/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2437 - accuracy: 0.1619\n",
            "Epoch 302/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2421 - accuracy: 0.1552\n",
            "Epoch 303/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2420 - accuracy: 0.1608\n",
            "Epoch 304/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2435 - accuracy: 0.1574\n",
            "Epoch 305/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2429 - accuracy: 0.1608\n",
            "Epoch 306/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2425 - accuracy: 0.1574\n",
            "Epoch 307/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2422 - accuracy: 0.1707\n",
            "Epoch 308/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2423 - accuracy: 0.1652\n",
            "Epoch 309/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2420 - accuracy: 0.1585\n",
            "Epoch 310/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2427 - accuracy: 0.1641\n",
            "Epoch 311/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2416 - accuracy: 0.1652\n",
            "Epoch 312/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2419 - accuracy: 0.1619\n",
            "Epoch 313/500\n",
            "29/29 [==============================] - 0s 8ms/step - loss: 2.2408 - accuracy: 0.1663\n",
            "Epoch 314/500\n",
            "29/29 [==============================] - 0s 8ms/step - loss: 2.2420 - accuracy: 0.1663\n",
            "Epoch 315/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2414 - accuracy: 0.1641\n",
            "Epoch 316/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2409 - accuracy: 0.1652\n",
            "Epoch 317/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2418 - accuracy: 0.1674\n",
            "Epoch 318/500\n",
            "29/29 [==============================] - 0s 11ms/step - loss: 2.2408 - accuracy: 0.1596\n",
            "Epoch 319/500\n",
            "29/29 [==============================] - 0s 10ms/step - loss: 2.2401 - accuracy: 0.1652\n",
            "Epoch 320/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2415 - accuracy: 0.1630\n",
            "Epoch 321/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2421 - accuracy: 0.1641\n",
            "Epoch 322/500\n",
            "29/29 [==============================] - 0s 12ms/step - loss: 2.2409 - accuracy: 0.1641\n",
            "Epoch 323/500\n",
            "29/29 [==============================] - 0s 12ms/step - loss: 2.2414 - accuracy: 0.1574\n",
            "Epoch 324/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2403 - accuracy: 0.1630\n",
            "Epoch 325/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2417 - accuracy: 0.1674\n",
            "Epoch 326/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2406 - accuracy: 0.1685\n",
            "Epoch 327/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2405 - accuracy: 0.1652\n",
            "Epoch 328/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2410 - accuracy: 0.1585\n",
            "Epoch 329/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2403 - accuracy: 0.1641\n",
            "Epoch 330/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2408 - accuracy: 0.1630\n",
            "Epoch 331/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2401 - accuracy: 0.1652\n",
            "Epoch 332/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2409 - accuracy: 0.1696\n",
            "Epoch 333/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2403 - accuracy: 0.1608\n",
            "Epoch 334/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2402 - accuracy: 0.1530\n",
            "Epoch 335/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2404 - accuracy: 0.1608\n",
            "Epoch 336/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2403 - accuracy: 0.1641\n",
            "Epoch 337/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2394 - accuracy: 0.1596\n",
            "Epoch 338/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2397 - accuracy: 0.1585\n",
            "Epoch 339/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2396 - accuracy: 0.1652\n",
            "Epoch 340/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2409 - accuracy: 0.1630\n",
            "Epoch 341/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2395 - accuracy: 0.1652\n",
            "Epoch 342/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2398 - accuracy: 0.1608\n",
            "Epoch 343/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2397 - accuracy: 0.1508\n",
            "Epoch 344/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2396 - accuracy: 0.1718\n",
            "Epoch 345/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2386 - accuracy: 0.1574\n",
            "Epoch 346/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2388 - accuracy: 0.1630\n",
            "Epoch 347/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2386 - accuracy: 0.1530\n",
            "Epoch 348/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2393 - accuracy: 0.1596\n",
            "Epoch 349/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2392 - accuracy: 0.1541\n",
            "Epoch 350/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2383 - accuracy: 0.1563\n",
            "Epoch 351/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2398 - accuracy: 0.1596\n",
            "Epoch 352/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2399 - accuracy: 0.1585\n",
            "Epoch 353/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2384 - accuracy: 0.1585\n",
            "Epoch 354/500\n",
            "29/29 [==============================] - 0s 7ms/step - loss: 2.2393 - accuracy: 0.1663\n",
            "Epoch 355/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2379 - accuracy: 0.1530\n",
            "Epoch 356/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2385 - accuracy: 0.1685\n",
            "Epoch 357/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2399 - accuracy: 0.1585\n",
            "Epoch 358/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2383 - accuracy: 0.1696\n",
            "Epoch 359/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2382 - accuracy: 0.1741\n",
            "Epoch 360/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2385 - accuracy: 0.1630\n",
            "Epoch 361/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2395 - accuracy: 0.1596\n",
            "Epoch 362/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2389 - accuracy: 0.1674\n",
            "Epoch 363/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2385 - accuracy: 0.1652\n",
            "Epoch 364/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2379 - accuracy: 0.1641\n",
            "Epoch 365/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2389 - accuracy: 0.1685\n",
            "Epoch 366/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2384 - accuracy: 0.1596\n",
            "Epoch 367/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2390 - accuracy: 0.1563\n",
            "Epoch 368/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2393 - accuracy: 0.1630\n",
            "Epoch 369/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2376 - accuracy: 0.1608\n",
            "Epoch 370/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2382 - accuracy: 0.1596\n",
            "Epoch 371/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2378 - accuracy: 0.1552\n",
            "Epoch 372/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2382 - accuracy: 0.1596\n",
            "Epoch 373/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2381 - accuracy: 0.1652\n",
            "Epoch 374/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2385 - accuracy: 0.1630\n",
            "Epoch 375/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2377 - accuracy: 0.1641\n",
            "Epoch 376/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2383 - accuracy: 0.1630\n",
            "Epoch 377/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2378 - accuracy: 0.1630\n",
            "Epoch 378/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2383 - accuracy: 0.1552\n",
            "Epoch 379/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2380 - accuracy: 0.1541\n",
            "Epoch 380/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2369 - accuracy: 0.1563\n",
            "Epoch 381/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2384 - accuracy: 0.1630\n",
            "Epoch 382/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2373 - accuracy: 0.1652\n",
            "Epoch 383/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2375 - accuracy: 0.1596\n",
            "Epoch 384/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2370 - accuracy: 0.1585\n",
            "Epoch 385/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2363 - accuracy: 0.1530\n",
            "Epoch 386/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2369 - accuracy: 0.1519\n",
            "Epoch 387/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2369 - accuracy: 0.1552\n",
            "Epoch 388/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2378 - accuracy: 0.1619\n",
            "Epoch 389/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2370 - accuracy: 0.1574\n",
            "Epoch 390/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2374 - accuracy: 0.1574\n",
            "Epoch 391/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2368 - accuracy: 0.1497\n",
            "Epoch 392/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2376 - accuracy: 0.1441\n",
            "Epoch 393/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2370 - accuracy: 0.1630\n",
            "Epoch 394/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2373 - accuracy: 0.1441\n",
            "Epoch 395/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2361 - accuracy: 0.1475\n",
            "Epoch 396/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2370 - accuracy: 0.1574\n",
            "Epoch 397/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2359 - accuracy: 0.1608\n",
            "Epoch 398/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2367 - accuracy: 0.1530\n",
            "Epoch 399/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2367 - accuracy: 0.1619\n",
            "Epoch 400/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2359 - accuracy: 0.1497\n",
            "Epoch 401/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2368 - accuracy: 0.1530\n",
            "Epoch 402/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2364 - accuracy: 0.1552\n",
            "Epoch 403/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2365 - accuracy: 0.1585\n",
            "Epoch 404/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2367 - accuracy: 0.1596\n",
            "Epoch 405/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2373 - accuracy: 0.1585\n",
            "Epoch 406/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2363 - accuracy: 0.1685\n",
            "Epoch 407/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2365 - accuracy: 0.1563\n",
            "Epoch 408/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2366 - accuracy: 0.1596\n",
            "Epoch 409/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2373 - accuracy: 0.1652\n",
            "Epoch 410/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2363 - accuracy: 0.1596\n",
            "Epoch 411/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2367 - accuracy: 0.1541\n",
            "Epoch 412/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2365 - accuracy: 0.1519\n",
            "Epoch 413/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2362 - accuracy: 0.1574\n",
            "Epoch 414/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2356 - accuracy: 0.1585\n",
            "Epoch 415/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2356 - accuracy: 0.1585\n",
            "Epoch 416/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2353 - accuracy: 0.1530\n",
            "Epoch 417/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2368 - accuracy: 0.1563\n",
            "Epoch 418/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2362 - accuracy: 0.1596\n",
            "Epoch 419/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2350 - accuracy: 0.1574\n",
            "Epoch 420/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2363 - accuracy: 0.1563\n",
            "Epoch 421/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2356 - accuracy: 0.1641\n",
            "Epoch 422/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2361 - accuracy: 0.1574\n",
            "Epoch 423/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2359 - accuracy: 0.1652\n",
            "Epoch 424/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2356 - accuracy: 0.1519\n",
            "Epoch 425/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2362 - accuracy: 0.1619\n",
            "Epoch 426/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2364 - accuracy: 0.1541\n",
            "Epoch 427/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2353 - accuracy: 0.1574\n",
            "Epoch 428/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2360 - accuracy: 0.1608\n",
            "Epoch 429/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2355 - accuracy: 0.1619\n",
            "Epoch 430/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2355 - accuracy: 0.1641\n",
            "Epoch 431/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2348 - accuracy: 0.1652\n",
            "Epoch 432/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2352 - accuracy: 0.1596\n",
            "Epoch 433/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2349 - accuracy: 0.1563\n",
            "Epoch 434/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2359 - accuracy: 0.1541\n",
            "Epoch 435/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2355 - accuracy: 0.1541\n",
            "Epoch 436/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2354 - accuracy: 0.1508\n",
            "Epoch 437/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2346 - accuracy: 0.1519\n",
            "Epoch 438/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2364 - accuracy: 0.1663\n",
            "Epoch 439/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2355 - accuracy: 0.1585\n",
            "Epoch 440/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2351 - accuracy: 0.1552\n",
            "Epoch 441/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2352 - accuracy: 0.1608\n",
            "Epoch 442/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2351 - accuracy: 0.1630\n",
            "Epoch 443/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2354 - accuracy: 0.1729\n",
            "Epoch 444/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2355 - accuracy: 0.1674\n",
            "Epoch 445/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2352 - accuracy: 0.1663\n",
            "Epoch 446/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2355 - accuracy: 0.1619\n",
            "Epoch 447/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2364 - accuracy: 0.1630\n",
            "Epoch 448/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2346 - accuracy: 0.1596\n",
            "Epoch 449/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2346 - accuracy: 0.1663\n",
            "Epoch 450/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2353 - accuracy: 0.1563\n",
            "Epoch 451/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2354 - accuracy: 0.1652\n",
            "Epoch 452/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2352 - accuracy: 0.1574\n",
            "Epoch 453/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2347 - accuracy: 0.1630\n",
            "Epoch 454/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2345 - accuracy: 0.1574\n",
            "Epoch 455/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2350 - accuracy: 0.1674\n",
            "Epoch 456/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2347 - accuracy: 0.1729\n",
            "Epoch 457/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2348 - accuracy: 0.1519\n",
            "Epoch 458/500\n",
            "29/29 [==============================] - 0s 7ms/step - loss: 2.2344 - accuracy: 0.1608\n",
            "Epoch 459/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2349 - accuracy: 0.1619\n",
            "Epoch 460/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2345 - accuracy: 0.1608\n",
            "Epoch 461/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2351 - accuracy: 0.1585\n",
            "Epoch 462/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2346 - accuracy: 0.1608\n",
            "Epoch 463/500\n",
            "29/29 [==============================] - 0s 6ms/step - loss: 2.2354 - accuracy: 0.1585\n",
            "Epoch 464/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2346 - accuracy: 0.1596\n",
            "Epoch 465/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2345 - accuracy: 0.1497\n",
            "Epoch 466/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2346 - accuracy: 0.1630\n",
            "Epoch 467/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2351 - accuracy: 0.1619\n",
            "Epoch 468/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2345 - accuracy: 0.1652\n",
            "Epoch 469/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2339 - accuracy: 0.1608\n",
            "Epoch 470/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2348 - accuracy: 0.1685\n",
            "Epoch 471/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2350 - accuracy: 0.1608\n",
            "Epoch 472/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2342 - accuracy: 0.1641\n",
            "Epoch 473/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2339 - accuracy: 0.1596\n",
            "Epoch 474/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2337 - accuracy: 0.1563\n",
            "Epoch 475/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2338 - accuracy: 0.1652\n",
            "Epoch 476/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2338 - accuracy: 0.1563\n",
            "Epoch 477/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2345 - accuracy: 0.1585\n",
            "Epoch 478/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2341 - accuracy: 0.1585\n",
            "Epoch 479/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2341 - accuracy: 0.1630\n",
            "Epoch 480/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2343 - accuracy: 0.1619\n",
            "Epoch 481/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2343 - accuracy: 0.1541\n",
            "Epoch 482/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2345 - accuracy: 0.1530\n",
            "Epoch 483/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2343 - accuracy: 0.1552\n",
            "Epoch 484/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2332 - accuracy: 0.1630\n",
            "Epoch 485/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2337 - accuracy: 0.1608\n",
            "Epoch 486/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2341 - accuracy: 0.1497\n",
            "Epoch 487/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2340 - accuracy: 0.1497\n",
            "Epoch 488/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2335 - accuracy: 0.1552\n",
            "Epoch 489/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2339 - accuracy: 0.1508\n",
            "Epoch 490/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2336 - accuracy: 0.1652\n",
            "Epoch 491/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2334 - accuracy: 0.1652\n",
            "Epoch 492/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2337 - accuracy: 0.1663\n",
            "Epoch 493/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2346 - accuracy: 0.1608\n",
            "Epoch 494/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2330 - accuracy: 0.1641\n",
            "Epoch 495/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2336 - accuracy: 0.1663\n",
            "Epoch 496/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2341 - accuracy: 0.1630\n",
            "Epoch 497/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2336 - accuracy: 0.1630\n",
            "Epoch 498/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2330 - accuracy: 0.1552\n",
            "Epoch 499/500\n",
            "29/29 [==============================] - 0s 5ms/step - loss: 2.2341 - accuracy: 0.1663\n",
            "Epoch 500/500\n",
            "29/29 [==============================] - 0s 4ms/step - loss: 2.2340 - accuracy: 0.1707\n"
          ]
        }
      ],
      "source": [
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "#-# STEP 6:  Train model  #-#\n",
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "\n",
        "if __name__ == '__main__':\n",
        "  epochs = 500\n",
        "  ### TO DO ###\n",
        "  model_cbow.fit(x_cbow_1hot, y_cbow_1hot, epochs = epochs)\n",
        "  model_skipgram.fit(x_skipgram_1hot, y_skipgram_1hot, epochs=epochs)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "G2855F0H8P4c"
      },
      "source": [
        "# **Section 4** Visualising the outcome (1 task to complete)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0qOql646S4fL"
      },
      "source": [
        "Now that the model has been trained, let's see if we have really achieved \"similar words placed together\".  First, create the following function:\n",
        "\n",
        "<center>\n",
        "\n",
        "| function name    | parameter name       | parameter datatype          | return type  |\n",
        "|------------------|----------------------|-----------------------------|--------------|\n",
        "| `get_embeddings` | `words`<br>`weights`<br>`word_to_index`<br>`num_vocabs` | `list[str]`<br>`np.ndarray`<br>`dict[str, int]`<br>`int` | `np.ndarray` |\n",
        "\n",
        "</center>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Ge6T9j_Ew1AC"
      },
      "outputs": [],
      "source": [
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "#-# STEP 7:  Extracting the embeddings  #-#\n",
        "#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#-#\n",
        "\n",
        "def get_embeddings(words, weights, word_to_index, num_vocabs):\n",
        "  \"\"\"\n",
        "  Takes in a list of strings and returns an array of embeddings corresponding to those strings.\n",
        "  Args:\n",
        "    words: a list of words in which we want to get the embedding for\n",
        "    weights: the embeddings learned by a model\n",
        "    word_to_index: the dictionary used to map the words back to integers\n",
        "    num_vocabs: the number of unique vocabs\n",
        "  Returns:\n",
        "    a 2D numpy array, where the n-th row corresponds to the embeddings of the n-th word in the list \"words\"\n",
        "  \"\"\"\n",
        "  ### TO DO ###\n",
        "  embeddings = np.zeros((len(words), weights.shape[1]))  # Initialize an array to store the embeddings\n",
        "\n",
        "  for i, word in enumerate(words):\n",
        "    if word in word_to_index:\n",
        "      index = word_to_index[word]  # Get the index of the word\n",
        "      embeddings[i] = weights[index]  # Retrieve the corresponding embedding\n",
        "  print(embeddings)\n",
        "  return embeddings"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UAJPShqXdjMC"
      },
      "source": [
        "Now, run the code in the following cell to visualise the semantic space!  You can try to extract different dimensions of the space to see what changes.  In the example code below, we extracted dimension 10 and dimension 20.  You could also try to use `weights_cbow` and see what changes."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "0kYUl3hiM61W",
        "outputId": "cf29176d-1890-4ef6-ab88-a017eb169eab"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[ 0.15062073 -0.89070827  0.7272808  ... -0.49018267 -0.52603233\n",
            "  -0.84332335]\n",
            " [ 0.25417173  0.04263366  0.88362771 ...  0.52005124  0.7269178\n",
            "   0.28514084]\n",
            " [ 0.02644332 -0.67982662  1.06070483 ... -0.23798041  0.84100389\n",
            "   0.79328549]\n",
            " ...\n",
            " [ 0.53820229 -0.66188478 -0.07772418 ... -0.65703058 -0.03228909\n",
            "   0.65920311]\n",
            " [ 0.04132812  0.83642167 -0.82438427 ...  0.4346723   0.79256052\n",
            "   0.53827274]\n",
            " [ 0.29334709  0.11489582  0.72274786 ... -0.02614549  0.81582481\n",
            "  -0.06242605]]\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 1500x1500 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABMkAAASuCAYAAADYoX52AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAAEAAElEQVR4nOzdeVhV5frG8XttBjcz4gSOoOCAI6bmLDkElh4tPZqZQzmUZmrpMT0d51KznNLqFJaYWg5pnkzDTCWHLDVnMUsCtcTIUBAVUTa/P/i5cwsomoC4v5/r2tdhr/WutZ61Kw7cvO+zjMzMzEwBAAAAAAAAdsxU2AUAAAAAAAAAhY2QDAAAAAAAAHaPkAwAAAAAAAB2j5AMAAAAAAAAdo+QDAAAAAAAAHaPkAwAAAAAAAB2j5AMAAAAAAAAds+xsAu42ywWi06dOiUPDw8ZhlHY5QAAAAAAAKAQZWZm6vz58ypbtqxMptzni913IdmpU6dUoUKFwi4DAAAAAAAA95CTJ0+qfPnyue6/70IyDw8PSVk37unpWcjVAAAAAAAAoDClpKSoQoUK1swoN/ddSHZtiaWnpychGQAAAAAAACTplm25aNwPAAAAAAAAu0dIBgAAAAAAALtHSAYAAAAAAAC7R0gGAAAAAAAAu0dIBgAAAAAAALtHSAYAAAAAAAC7R0gGAAAAAAAAu0dIBgAAAAAAALtHSAYAAAAAAAC7R0gGAAAAAAAAu0dIBgAAAAAAALtHSAYAAAAAAAC7R0gGAAAAAAAAu0dIBgAAAAAAALtHSAYAAAAAAAC7R0gGAAAAAAAAu5evIdmWLVvUsWNHlS1bVoZhaPXq1TcdHx0dLcMwsr1Onz6dn2UCAAAAAADAzuVrSHbhwgXVrVtXb7/99m0dd/ToUSUkJFhfpUuXzqcKAQAAAAAAAMkxP0/evn17tW/f/raPK126tLy9vfM09vLly7p8+bL1fUpKym1fDwAAAAAAAPbtnuxJVq9ePfn5+aldu3bavn37TcdOnTpVXl5e1leFChUKqEoAAAAAAADcL+6pkMzPz0///e9/tXLlSq1cuVIVKlRQaGio9uzZk+sxY8aMUXJysvV18uTJAqwYAAAAAAAA94N8XW55u6pVq6Zq1apZ3zdt2lSxsbGaNWuWFi1alOMxxYoVU7FixQqqRAAAAAAAANyH7qmZZDlp1KiRjh07VthlAAAAAAAA4D52z4dk+/btk5+fX2GXAQAAAAAAgPtYvi63TE1NtZkFFhcXp3379snHx0cVK1bUmDFj9Ntvv+mjjz6SJM2ePVsBAQGqWbOm0tLSNH/+fG3atElfffVVfpYJAAAAAAAAO5evIdnu3bv10EMPWd+/9NJLkqQ+ffooMjJSCQkJOnHihHV/enq6RowYod9++02urq6qU6eOvv76a5tzAAAAAAAAAHebkZmZmVnYRdxNKSkp8vLyUnJysjw9PQu7HAAAAAAAABSivGZF93xPMgAAAAAAACC/EZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAIAip1q1ajIMo7DLAAAAwH2EkAwAANwxwzDk4uJS2GUAAAAAfxshGQAAAAAAAOweIRkAALgjTk5OkqS0tDQZhiHDMPT222/r8ccft743DEPu7u46d+6c9TjDMGQ2m2U2m61jvLy8dPXqVeuYLVu2yMnJybrfZDJp2rRpN62nUqVKNtcNDg622T9x4kSb/Y888ogMw1CTJk0kSSaTSWaz2eaYL774QoZh6NFHH/07HxUAAACKAEIyAABwR7Zs2SIpKyxbsWKFVqxYIQ8PD3322WcqVqyYJkyYoMaNG+vChQvy9/e3Ofby5csyDEOTJk1SzZo1lZKSoqpVq1r3P/TQQ7p69aq6du2qYcOGSZLGjBmjM2fO5FjLgw8+qBMnTqhChQp69dVXVaZMGR05ckSdOnWSJKWmpmrChAkyDEPDhg1Tt27d9OWXX9qco2bNmrp8+bJ+/fVX67YBAwZIkv73v//9vQ8LAAAA9zwjMzMzs7CLuJtSUlLk5eWl5ORkeXp6FnY5AADcd1LTrurFZXt14uwlffViK5nNZl26dEmS5OHhodTUVF25ckWOjo6SpIoVK+rkyZO6dOmSdfaYJJsx7u7uunDhgjIzMzVt2jSNGTNGTz75pJYsWSJJ2rBhgx5++GHVr19fP/zwg6pVq6affvpJ136MMQxDTk5OSk9Pt9bp6Ogoi8Uii8Witm3bauPGjVqxYoW6du0qSXrkkUf05ZdfqnHjxtqxY4fi4+MVEBCgkJAQ7dmzx3peb29vnT17tgA+WQAAAOSHvGZFzCQDAAB59o95W1VrwnptOJKoo6fPS5LSrlj0j3lbJUkXL16Uo6OjNfySpMcff1yS9NFHH1m3OTk52YypU6eOpKwZX2vWrJEkvffee9b97dq1kyT98ssvudYWEBBg87506dLWEC0mJkaSrAGZlLX88nr+/v4ym83at2+fJGnw4MGSpKlTp+Z6TQAAANw/CMkAAECe/GPeVh34NSXHfQd+TbEGZUXZkCFDlJmZqcjISC1YsECGYei5554r7LIAAABQAAjJAADALaWmXc01IJOyZmsd+DVFrq6uunr1qk0T/lWrVkmSevfubd125coVmzMcPHhQUtayy44dO0qSnn32Wev+DRs2SJIqV66ca41xcXE27xMTE61LO6818b9WiyRNmDAh2zneeOMNSdKwYcOUlpamWrVq5Xo9AAAA3F8IyQAAwC29uGxvzjtMjlLGFaX88IUuxv6g+j1GSMoKuyZOnKimTZvq5MmT8vLyyvbkSFdXV02ePFl16tRRamqqKlWqJEkaPXq0TCaTPv74Y3Xv3l3Dhw9XWFiYJGn9+vU5ltG4cWNduXJFlSpV0rRp0+Tr66uMjAxr4/7Vq1dLylpuOWLECPXo0UPr1q2TJGuQdk1AQIBSUrICwc8///w2PykAAAAUVTTuBwAAtxQ2e4u1B9n1zn27TMlbF1nfV+8yTDUsJ/TZZ59Zt7m5uenXX3+Vt7e3pKxQqlixYsrMzLQ22nd3d9fZs2etfcq2bNmiNm3aWGekGYahKVOmaPTo0ZKUrXG/JFWqVEknTpywvq9Ro4a1F5mU1YPs+tljLVq00NatW9W2bVvrTDVJOnDggOrWratixYopLS3ttj8rAAAA3FvymhU55roHAADg/1Us7pJjSObdtLu8m3a3vm9eo7Qi+jS85fkMw7hpANWyZctsSzKvd/To0Wzbjh8/ftNrjh8/XuPHj7e+79GjhyQpPDzcZtzWrVm91Z555pmbng8AAAD3F2aSAQCAW0pNu6paE3Je6ni9QxPC5G6++d/gDMOQ2WzWpUuX7lZ5efLggw+qePHi6tixo9auXasvv/xSkqyz0c6cOaPo6Gg98cQTysjI0H32IxIAAIDdYiYZAAC4a9zNjqpT3vMmzfulOuU9bxmQFabk5GTt3LnT2tfM0dHR2pdMkgYMGGDtXTZy5MjCKBEAAACFiJlkAAAgz/4xb2uOQVmd8p76fEiLQqgIAAAAuDlmkgEAgLvu8yEtlJp2VS8u26sTZy+pYnEXzeoeck/PIAMAAADygp9oAQDAbXE3O+apOT8AAABQlJgKuwAAAAAAAACgsBGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO7la0i2ZcsWdezYUWXLlpVhGFq9evUtj4mOjlb9+vVVrFgxBQYGKjIyMj9LBAAAAAAAAPI3JLtw4YLq1q2rt99+O0/j4+Li9Oijj+qhhx7Svn37NHz4cPXv31/r16/PzzIBAAAAAABg5xzz8+Tt27dX+/bt8zz+v//9rwICAjRjxgxJUo0aNbRt2zbNmjVLYWFh+VUmAAAAAAAA7Nw91ZNsx44datu2rc22sLAw7dixI9djLl++rJSUFJsXAAAAAAAAcDvuqZDs9OnTKlOmjM22MmXKKCUlRZcuXcrxmKlTp8rLy8v6qlChQkGUCgAAAAAAgPvIPRWS3YkxY8YoOTnZ+jp58mRhlwQAAAAAAIAiJl97kt0uX19f/f777zbbfv/9d3l6esrFxSXHY4oVK6ZixYoVRHkAAAAAAAC4T91TM8maNGmijRs32mzbsGGDmjRpUkgVAQAAAAAAwB7ka0iWmpqqffv2ad++fZKkuLg47du3TydOnJCUtVSyd+/e1vHPPfecfvnlF40aNUo//vij3nnnHS1fvlwvvvhifpYJAAAAAAAAO5evIdnu3bsVEhKikJAQSdJLL72kkJAQjRs3TpKUkJBgDcwkKSAgQGvXrtWGDRtUt25dzZgxQ/Pnz1dYWFh+lgkAAAAAAAA7Z2RmZmYWdhF3U0pKiry8vJScnCxPT8/CLgcAAAAAAACFKK9Z0T3VkwwAAAAAAAAoDIRkAAAAAAAAsHuEZAAAAAAAALB7hGQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAAALB7hGQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAAALB7hGQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAAALB7hGQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAAALB7hGQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAAALB7hGQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAAALB7hGQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAAALB7hGQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAAALB7hGQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAAALB7hGQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAAALB7hGQAAOQTR0dHlSpVKs/ju3TpIsMwZBiGSpUqpZYtW8owjHysMMuQIUNkGIYOHDiQ79cCAAAA7lWOhV0AAADIsmrVKvn5+WnBggUKCAiQm5ubYmNjC7ssAAAAwC4QkgEAcA+4FoZ17txZYWFh1u3lypUrrJIAAAAAu8JySwAA7oK4uDh5eHhYl0sGBwfb7P/jjz/k6+tr3e/g4KAhQ4ZIylruGBgYKEl69913ZRiGhgwZkm25pb+/v0wmk1q3bm09j6urq3766SfrmPT0dAUFBVn3m0wmdezY0aaWnj17Wvc7OTmxzBJ3xezZs2UYho4fP17YpQAAANwRQjIAAO6CBx54QKmpqRowYIBmzJih2NhYZWRkWPcHBQXpjz/+0LBhw7Ro0SJVrVpVb7/9tiIjIzVp0iTNnz9fktSpUydt3LhRkyZNyvE6mZmZ2rFjh2bNmqVhw4bp0qVLat26tXV/cHCwjh07pj59+mjJkiVq3ry5vvjiC73wwguSpM8//1wff/yxSpcurffff1+tWrXS1q1b8/GTAQAAAIoGllsCAPA3xcbG6uzZs3r00Uf1/vvvS5JatWqlBg0aSJI+++wzJScnKyoqyrqU8qmnnpKTk5P+85//qG/fvmrYsKEkqXz58jahV07279+vqlWrSpKWLVumhIQESVmz1WJjYzVq1Ci9/vrrkqQnn3xSXl5e+vDDDzV37lwNHjxYhmHo999/lyQNGDBAZcuWtZ4DAAAAsFfMJAMA4A5cSs/Q2NUH1euD7zVw4tuSpH79+ln3P/DAA9alkitWrJAkhYeHW5c5Goahq1evKikp6bauaxiGNSCTpJIlS8pisUiSli5dKkmaPn26zXVSUlKUlpYmSTpz5ow8PDxsztm0adPbqgFF24gRI2Q2m63LcX18fJSYmChvb2+FhITYjPXz87MuBZaklJQUPfjgg3J0dLQu1+3bt6/NMcuXL5ebm5sMw5CHh4e+/PLLgrgtAACAv42ZZAAA3KYBH+3ShphE6/vkk+ckSW9HH9Njj2Uff/bsWUnS+++/r2LFitnsK1++/F2r6/Tp05Kyepxdm5l2jbe39127Doquffv2aebMmerUqZNGjhyp33//XStWrLAGrbdSq1YtnTp1SiNGjFD79u21a9cu/frrrzZjJk6cqClTpiggIEC9evVS9+7dlZKSkh+3AwAAcFcRkgEAcBtuDMgkySWoic5FL9D2L1ZowAOhiujdUPv27VNmZqYkqWvXroqKitLhw4c1e/bsfKutT58+mjJlig4fPqy5c+fmOKZkyZI6deqUzbYdO3bkW00ofOlXLVq0I17Hky4qYfdmSdK//vUvNWvWTJLUpUuXPJ1n/fr1OnnypF5//XWNGjVKkhQaGppt3Pjx4zVs2DBJ0ksvvaQJEybo3LlzBLUAAOCex3JLAADy6FJ6RraATJKcfcrKKOamtF926ZNZ4/TGjFlq3LixdX+/fv3k5uamOXPmqHPnzlq2bJnGjh2roKAgPfnkk3etvqpVq6p8+fLavHmzWrVqpcWLF+vVV19VvXr11KpVK0nS3LlzlZmZKV9fX0VERKhdu3bZQjPcP6aui1H1sV9q8toj+mjHca2/5C+T2V3NmzdX+fLl1bt3b8XFxeXpXFFRUZJkfSprbjp06GD9+trS4CNHjtzhHQAAABQcZpIBAJBHU9bF5LrPt/csnY4cqgv7ojRqX5SqV6+un3/+2br/t99+U506dfS///1P//vf/yRJxYoV06BBg+5qjfHx8QoJCdGWLVu0ZcsWSZKTk5P1Oo899pi6d++uZcuWaeDAgXJwcFDz5s21bdu2u1oHCt/UdTF6b4ttAGZydFa5Fz5W6r4oOSV8rxUrVmjx4sWKjo6WYRjW2Y/XXP+EVnd39zxd19XV9a/rmUzZzgMAAHCvIiQDACCP4v+8mOs+Z5+yqvjSp5KkFkEltajfgzb7vby8dPz48VyPr1OnTraA4lrIZb1+fHy24w4ePGjz3sHBQQcOHMj1OlJWg/9rTf5xf0q/alHE1pxniJlMJnnWf0SG8Yj+GNtGXh5umjFjhjw8PPTnn3/+dY70dCUlJcnT01OSFBYWpldffVXz5s2zLrcEAAC4nxCSAQCQR/4lXLX157yNAwrToh3xsmRm335+/1e6+ONWuQWHytGnrB5/epUsFoseeOABubu76+OPP9aECRPUqFEjjRgxwmYGWPPmzRUYGKh///vfOnv2rNq3b689e/bo+PHjmjVrVgHeHQAAQP4gJAMAII/+/UiwFn13Ik/jgMJ0PCnnWY8Obt5K/z1WafH7JGVqo4OjunbtqnHjxunixYvav3+/Jk2aJEl6+OGHde7cOZvjd+3apTZt2mj69OmaNm2anJyc9NRTT+XvzQAA7lh0dLQeeughnT17lgeoAHlgZN64tqOIS0lJkZeXl5KTk63LAwAAuFtyerrl9doFl1ZE74YFWBGQ3Qdbf9Hktbdulj/20Rrq16JyAVQEACgMhGRAlrxmRTzdEgCA2xDRu6HaBZfOcR8BGe4VvZr4y2TcfIzJyBoHAMDtSk9PL+wSgHxBSAYAwG2K6N1QRyaFq1fjimoRVFK9GlfUkUnhBGS4Zzg7mjSgRcBNxwxoESBnR34UBID8EBoaqhdeeEHDhw9X8eLFVaZMGUVEROjChQt6+umn5eHhocDAQH355ZfWYw4dOqT27dvL3d1dZcqUUa9evXTmzJm/dc5rtm/frjp16shsNqtx48Y6dOiQzf5t27apRYsWcnFxUYUKFTR06FBduHDBut/f31+TJ09W79695enpqYEDB+bDpwYUPn4yAgDgDrg4O2hy59pa1O9BTe5cWy7ODoVdEmBjzCPBerZlQLYZZSZDerZlgMbQOw+4pxmGodq1axd2GfgbFi5cqJIlS2rnzp164YUXNGjQIP3zn/9U06ZNtWfPHj388MPq1auXLl68qHPnzql169YKCQnR7t27FRUVpd9//13dunW743Ne71//+pdmzJihXbt2qVSpUurYsaOuXLkiSYqNjVV4eLi6dOmiAwcOaNmyZdq2bZuGDBlic44333xTdevW1d69ezV27Nj8/fCAQkJPMgAAgPtY+lWLFu2I1/Gki6rk46peTfyZQQYUAdu3b1eFChVUsWLFwi4FeZBhydTOuCQlnk9TaQ+zRvfrIktGhrZu3Zq1PyNDXl5eevzxx/XRRx9Jkk6fPi0/Pz/t2LFDX3/9tbZu3ar169dbz/nrr7+qQoUKOnr0qKpWrarQ0FBl3MY5GzdubO1JtnTpUnXv3l2SlJSUpPLlyysyMlLdunVT//795eDgoPfee8967W3btqlVq1a6cOGCzGaz/P39FRISos8++6xAPk/gbstrVsTTLQEAAO5jzo4mmvMDRVCzZs0KuwTkUdShBE1cE6OE5DTrtqQTZ9WqUYj1vYODg0qUKGEzO7BMmTKSpMTERO3fv1+bN2+Wu7t7tvPHxsaqatWqkqQ6derk+ZzXa9KkifVrHx8fVatWTUeOZD3gZf/+/Tpw4ICWLFliHZOZmSmLxaK4uDjVqFFDktSgQYO8fiRAkcWfEQEAQL4yDENjxowp7DIA4J7y008/yc3NTYZhWJdWOjo6qlSpUpJsl1u6ubnJ1dXV5vjk5GQZhqHQ0FBJWY3Ug4KCrOczmUzq2LGjdfyQIUNkGIYGDhwoBwcHGYYhBwcHRUREFMwN36eiDiVo0OI9NgGZlDWL95tjZxV1KMG6zTAMOTk52byXJIvFotTUVHXs2FH79u2zef38889q2bKl9Zjrj7/VOfMqNTVVzz77rM119+/fr59//llVqlSxjnNzc8vzOYGiipAMAADcFaGhoXJxccm2ff/+/YRkAHCDRo0a6eLFixowYIBmzZqln3/+WRkZGTmO7dSpky5duqTY2FjrtmeffVaS9P7770uSgoODdezYMfXp00dLlixR8+bN9cUXX+iFF16wOdcHH3ygwYMHa+7cuTIMQ88991w+3eH9L8OSqYlrYnSz/kUT18Qow3LrDkf169fX4cOH5e/vr8DAQJvX3QinvvvuO+vXZ8+e1U8//WSdIVa/fn3FxMRku25gYKCcnZ3/9rWBooSQDAAA5Ks6derQJxSA3buUnqGxqw+q1wffa/Dba5ScnKwOHTro/fff1/Dhw7V79+5cj70WhA0YMMC6bc2aNXJxcVHVqlX1xx9/KDY2VqNGjVJkZKSefPJJbdmyRZ6envrwww9tztWvXz/NnTtXQ4YMUffu3WWxWHT69On8uen73M64pGwzyG6UkJymnXFJtzzX888/r6SkJPXo0UO7du1SbGys1q9fr6effjrX8PR2TJo0SRs3btShQ4fUt29flSxZUp07d5Ykvfzyy/r22281ZMgQ6+y1//3vf9ka9wP2gJAMAABIkry9vVW3bl01atRIJpNJDg4O1mU8krRjxw75+flZl/JUqFDB+gj5/v3765tvvlFaWpp1f//+/SVlX265cuVKFS9e3LocqEaNGja/oAUGBsrPz08dOnSQg4ODTCaTateune1JXQBQVAz4aJdqjIvSou9OaOvPZ/TJp1nNzy9V/msZXa1ataxL5W7k7u4uHx8fbdu2TZIUFxenixcvWpdTLl26VJI0ffp06/dgwzCUkpKitDTbEKdPnz4215Syvr/j9iWev3lAdjvjypYtq+3btysjI0MPP/ywateureHDh8vb21sm09//tX3atGkaNmyYHnjgAZ0+fVpr1qyxzhKrU6eOvvnmG/30009q0aKFQkJCNG7cOJUtW/ZvXxcoamjcDwAArA4cOKDQ0FCtX79eK1asUEREhKZNm6aRI0eqTZs2cnJy0qJFi3T58mUNHTpUzZs317lz5zR9+nTt27dPhw4d0s6dOyVJ5cuXz3b+xMREdevWTb6+vvrwww/1yy+/aPTo0WrevLmOHTtmHXf69GnFxcVpzZo1+u677zR58mQ999xz1qd3AUBRMeCjXdoQk5jjvv0nz2nAR7sU0bvhLc/Tr18/vfHGG9qyZYsmTZok6a8ZZtf+0DBkyBA1bGh7Lm9vb5v3Hh4e1q+vhS9Xr17N283ARmkPc677fJ+clm1cfHx8tnGZmX8txQwKCtKqVatyPWd0dHS2bbc6Z2hoqPV9hw4dcj13w4YN9dVXX+W6P6frAPcjQjIAAOxU+lWLFu2I1/Gki6rkk9UQ2tPTU5s3b5YktWvXTh9//LFWr14tSbp06ZI2b96sBx98UJJUqlQpderUSQsXLlSfPn3k7u4uwzBsnrx1o5EjRyozM1N79+5V6dKlJWU1DJ4wYYIOHTpkndXg4OCgvXv3ytnZWY888ogWLFiQ4y8HAHAvu5SekWNA5hL4oM5FL1Dqga+1oWpTXUrPUOxPR2zCjRtNnz5db7zxhoYOHaojR46oePHi8vLykpQ1O2zKlCk6fPiw5s6dm2/3A1uNAnzk52XW6eS0HPuSGZJ8vcxqFOBT0KUBuEMstwQAwA5NXRej6mO/1OS1R/TRjuOavPaIUtKuyFy8tM04Dw8PJSUladeuXXJ0dLQGZJL0j3/8Q4ZhaPv27Xm+7uHDh+Xl5WUNyCTp6aefliRrOCdlPZ7++mbBJUqU0Pnz52/7PgGgME1ZF5PjducS5WUUc1Va7E79uf5tdR78inUGWG5LLqWsGboHDhxQenq6+vbta91etWpVlS9fXps3b1arVq20ePFivfrqq6pXr55atWp1V+8Jf3EwGRrfMVhSViB2vWvvx3cMloMp93+mAO4thGQAANiZqeti9N6WOOX0sK2zlyyaesMvdbfzGPm7xcHBwea9YRg3nWEBAPei+D9z76Xo22umDMdiSt33pb764HUFBgbKZDLJyckp12P+/e9/W78XvvHGG7bXio9X7dq1tWXLFvXq1Utjx45VTEyM6tWrd1fuBTkLr+Wnd5+qL18v26WXvl5mvftUfYXX8iukygDcCZZbAgBgR9KvWhSxNe6mYyK2xmnEw9Xl7PjX39IaNmyoVatW6fvvv7fOJvv888+VmZmp5s2bS5KcnZ1vGWTVrFlTe/fuVWJionU22YIFCyRJDz300B3fFwDci/xLuGrrzznvcy5RXhVHrJQk9WpcUc/UdVflypXl7+8vSTl+Px00aJAGDRqU4/kcHBx04MCBXGuZN2+e5s2bZ7Pt5Zdf1ssvv5yHO8HNhNfyU7tgX+2MS1Li+TSV9shaYskMMqDoYSYZAAB2ZNGO+BxnkF3Pkpk17nqjRo2Si4uLHn74YS1ZskQLFixQjx495OXlpd69e0vKajh8+fJlLVu2TEePHlVKSkq2c8+cOVOGYah+/fr67LPPNHPmTL366quqUqWKtR8ZANwv/v1IcK77kncs1+/Lxur8oU1y2L9aNWrUkCTNmDGjoMrDXeRgMtSkSgl1qldOTaqUICADiihCMgAA7MjxpNyX/txsnMlk0saNG+Xq6qqnnnpKzzzzjHx8fLRt2zbrmKlTp6pcuXLq0aOHqlevrpdeeinbeUuWLKnly5frwoULevzxxzVy5EgFBgbanAcA7hcuzg5qF1w61/1p8fuUtHamJk4Yp4yMDM2YMUONGzcuwAoBANczMu+zBh8pKSny8vJScnKyPD09C7scAADuKR9s/UWT1x655bixj9ZQvxaVC6AiALj/DfhoV45PuWwXXFoRvRsWQkUAYF/ymhXRkwwAADvSq4m/Xlt35KZLLk1G1jgAwN0R0buhLqVnaMq6GMX/eVH+JVz170eC5eLscOuDAQAFhpAMAAA74uxo0oAWAXpvS+7N+we0CLBp2g8A+PtcnB00uXPtwi4DAHAThGQAANiZMf/fSDpia5zNjDKTkRWQjblJo2kAAADgfkVPMgAA7FT6VYsW7YjX8aSLquTjql5N/JlBBgAAgPsOPckAAMBNOTuaaM4PAAAA/D/+XAwAAAAAAAC7R0gGAAAAAAAAu0dIBgAAAAAAALtHSAYAAAAAAAC7R0gGAAAAAEVAaGiohg8fXthlAMB9i6dbAgAAAEAB8vf31/Dhw2878Fq1apWcnJzypygAACEZAAAAABQFPj4+hV0CANzXWG4JAAAAANexWCyaPn26AgMDVaxYMVWsWFGvvfaaJOngwYNq3bq1XFxcVKJECQ0cOFCpqanWY/v27avOnTvrzTfflJ+fn0qUKKHnn39eV65ckZS1ZPL48eN68cUXZRiGDMOQJP3555/q0aOHypUrJ1dXV9WuXVuffPKJTV03Lrf09/fXlClT9Mwzz8jDw0MVK1bU+++/b92fnp6uIUOGyM/PT2azWZUqVdLUqVPz62MDgCKPkAwAAAAArjNmzBhNmzZNY8eOVUxMjD7++GOVKVNGFy5cUFhYmIoXL65du3ZpxYoV+vrrrzVkyBCb4zdv3qzY2Fht3rxZCxcuVGRkpCIjIyVlLZksX768Jk2apISEBCUkJEiS0tLS9MADD2jt2rU6dOiQBg4cqF69emnnzp03rXXGjBlq0KCB9u7dq8GDB2vQoEE6evSoJOmtt97S559/ruXLl+vo0aNasmSJ/P397/rnBQD3CyMzMzOzsIu4m1JSUuTl5aXk5GR5enoWdjkAAAAAipDz58+rVKlSmjdvnvr372+zLyIiQi+//LJOnjwpNzc3SdK6devUsWNHnTp1SmXKlFHfvn0VHR2t2NhYOTg4SJK6desmk8mkpUuXSsp7T7IOHTqoevXqevPNNyVlzSSrV6+eZs+ebT1PixYttGjRIklSZmamfH19NXHiRD333HMaOnSoDh8+rK+//to6Yw0A7FFesyJmkgGFID4+XoZhaN++fYVdCgAAACRlWDK1I/ZP/fd/3+jy5csKfah1tjFHjhxR3bp1rQGZJDVr1kwWi8U6e0uSatasaQ3IJMnPz0+JiYk3v35GhiZPnqzatWvLx8dH7u7uWr9+vU6cOHHT4+rUqWP92jAM+fr6Wq/Vt29f7du3T9WqVdPQoUP11Vdf3fxDAAA7R0gGAAAAwK5FHUpQ89c3qUfEd5q1+bgk6Z///VZRhxLu6Hw3PoHSMAxZLJabHvPGG29ozpw5evnll7V582bt27dPYWFhSk9Pv+Nr1a9fX3FxcZo8ebIuXbqkbt26qWvXrndwRwBgHwjJgNt0qx9UAAAAUHREHUrQoMV7lJCcJklyKl5WhmMxnTy0U4MW77EJymrUqKH9+/frwoUL1m3bt2+XyWRStWrV8nxNZ2dnZWRk2Gzbvn27OnXqpKeeekp169ZV5cqV9dNPP/3Nu5M8PT3VvXt3RUREaNmyZVq5cqWSkpL+9nkB4H5ESAbcQmhoqIYMGaLhw4erZMmSCgsL06FDh9S+fXu5u7urTJky6tWrl86cOWM9JioqSs2bN5e3t7dKlCihDh06KDY2thDvAgAAADfKsGRq4poYXd+k2XB0lueDXXQ2eoFSD23UmMivtf3bHfrggw/Us2dPmc1m9enTR4cOHdLmzZv1wgsvqFevXipTpkyer+vv768tW7bot99+s/4MGRQUpA0bNujbb7/VkSNH9Oyzz+r333//W/c3c+ZMffLJJ/rxxx/1008/acWKFfL19ZW3t/ffOi8A3K8IyYA8WLhwoZydnbV9+3ZNmzZNrVu3VkhIiHbv3q2oqCj9/vvv6tatm3X8hQsX9NJLL2n37t3auHGjTCaTHnvssVtOswcAAEDB2RmXZJ1Bdj2vZk/Is+FjOrt1ifbNekZd/tlNiYmJcnV11fr165WUlKSGDRuqa9euatOmjebNm3db1500aZLi4+NVpUoVlSpVSpL0n//8R/Xr11dYWJhCQ0Pl6+urzp07/6378/Dw0PTp09WgQQM1bNhQ8fHxWrdunUwmfg0EgJzwdEsgBxmWTO2MS1Li+TRNfLabdOWS9uzZI0l69dVXtXXrVq1fv946/tdff1WFChV09OhRVa1aNdv5zpw5o1KlSungwYOqVauW4uPjFRAQoL1796pevXoFdVsAAAC4zv/2/aZhS/fdctycJ+qpU71y+V8QACBf5DUrcizAmoAiIepQgiauibH+VfF0Qoq8ylRQ1KEEhdfy0/79+7V582a5u7tnOzY2NlZVq1bVzz//rHHjxun777/XmTNnrDPITpw4oVq1ahXo/QAAACBnpT3Md3UcAKBoIyQDrnOtceuN0ysvZTpp0OI9evep+kpNTVXHjh31+uuvZzvez89PktSxY0dVqlRJERERKlu2rCwWi2rVqkXTfwAAgHtIowAf+XmZdTo5LdvPf5JkSPL1MqtRgE9BlwYAKASEZMD/y6lx640mrolRaEiIPlu1Sv7+/nJ0zP6f0J9//qmjR48qIiJCLVq0kCRt27Ytn6oGAADAnXIwGRrfMViDFu+RIdk28P///x3fMVgOJiOHowEA9xs6NgL/L7fGrddkSkpITlOTDk8qKSlJPXr00K5duxQbG6v169fr6aefVkZGhooXL64SJUro/fff17Fjx7Rp0ya99NJLBXcjAAAAyLPwWn5696n68vWyXVLp62XWu0/VV3gtv0KqDABQ0JhJBvy/xPO5B2TXy3Qtru3bt+vll1/Www8/rMuXL6tSpUoKDw+XyWSSYRhaunSphg4dqlq1aqlatWp66623FBoamr83AAAAgDsSXstP7YJ9rQ9uKu2RtcSSGWQAYF94uiXw/3bE/qkeEd/dctwnAxqrSZUSBVARAAAAAAD4u/KaFbHcEvh/1xq35vb3QkOSH41bAQAAAAC4LxGSAf/vWuNWSdmCMhq3AgAAAABwfyMkA65D41YAAADkxeXLlzV06FCVLl1aZrNZzZs3165duyRJ0dHRMgxDGzduVIMGDeTq6qqmTZvq6NGjhVw1AOBm6EkG5CDDkknjVgAAAORq2LBh+vTTTzV//nxVqlRJ06dP1+eff65jx47pwIEDeuihh/Tggw/q9ddfV6lSpfTcc88pIyND27dvL+zSAcDu5DUrIiQDAAAAgFu4/o+oHg4ZCn8gUJGRkXryySclSVeuXJG/v7+GDx+uhg0b6qGHHtLXX3+tNm3aSJLWrVunRx99VJcuXZLZbL7ZpQAAd1lesyLHAqwJAAAAAIqcqEMJmrgmRgnJaZKk9MQ4XblyRVdKBlnHODk5qVGjRjpy5IgaNmwoSapTp451v59fVtuOxMREVaxYsQCrBwDkFT3JAAAAACAXUYcSNGjxHmtAdr3/fHZIUYcScj3WycnJ+rVhZLXusFgsd79IAMBdQUgGAAAAADnIsGRq4poY3difxtHbT3JwVNpvMZq4JkYZlkxduXJFu3btUnBwcKHUCgD4+1huCQAAAAA52BmXlOMMMpOzWR71HtHZzR/qF7OHln3loq+Wvq+LFy+qX79+2r9/fyFUCwD4u5hJBgAAAAA5SDyfPSC7pnhoX7lWa6YzX8xQ306tdezYMa1fv17FixcvwAoBAHcTT7cEAAAAgBzsiP1TPSK+u+W4TwY0VpMqJQqgIgDAnchrVsRMMgAAAADIQaMAH/l5mWXkst+Q5OdlVqMAn4IsCwCQTwjJAAAAACAHDiZD4ztmNeK/MSi79n58x2A5mHKL0QAARQkhGQAAAADkIryWn959qr58vcw22329zHr3qfoKr+VXSJUBAO42nm4JAAAAADcRXstP7YJ9tTMuSYnn01TaI2uJJTPIAOD+QkgGAAAAALfgYDJozg8A9zmWWwIAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAAAAAAMDuEZIBAAAAAADA7hGSAQAAAAAAwO4RkgEAABRh/v7+mj17dmGXAQAAUOQRkgEAACBXkZGR8vb2LuwyAAAA8h0hGQAAAAAAAOweIRkAAEAh+/TTT1W7dm25uLioRIkSatu2rS5cuKDQ0FANHz7cZmznzp3Vt29fm23nz59Xjx495ObmpnLlyuntt9+27svMzNSECRNUrFgxGYahsmXLaujQodb9ly9f1siRI1WuXDm5ubnpwQcfVHR0tCQpOjpaTz/9tJKTk2UYhgzD0IQJEyRJ77zzjoKCgmQ2m1WmTBl17do1Pz4aAACAAuNY2AUAQEELDQ1VvXr16OED4J6QkJCgHj16aPr06Xrsscd0/vx5bd26VZmZmbc81tvbWxcvXtQbb7yhf//735o4caLWr1+vYcOGqWrVqmrXrp1WrlypWbNm6cMPP1T58uXl4uKi/fv3W88xZMgQxcTEaOnSpSpbtqw+++wzhYeH6+DBg2ratKlmz56tcePG6ejRo5Ikd3d37d69W0OHDtWiRYvUtGlTJSUlaevWrfn2GQEAABQEQjIAAIAClmHJ1M64JCWeT9PZ40d19epVPf7446pUqZIkqXbt2rd1vmbNmmn06NGSpKpVq2r79u2aNWuW2rVrp/j4eJUpU0bdunWTk5OTJKlRo0aSpBMnTmjBggU6ceKEypYtK0kaOXKkoqKitGDBAk2ZMkVeXl4yDEO+vr7W6504cUJubm7q0KGDPDw8VKlSJYWEhPztzwUAAKAwEZIBAAAUAG9vb1WsWFEXrlj0y4+HJRlyDQ6Vd2hfGY7O8vf3l2EY6tixoyIjI1W8eHGdOXNG7733nubMmSPDMOTv76/KlSvL29tbgYGBSk5OliRFRUXJMAxt3bpVu3fv1rJly+Tt7S1XV1ddunRJ3t7ecnd3V0ZGhlasWKGOHTvK0dFRzzzzjDIyMlSuXDlJkmEYcnV11eXLl1WiRIlc76Vdu3aqVKmSKleurPDwcIWHh+uxxx6Tq6trQXyUAAAA+YKeZADs2s168UjS8ePH1bFjRxUvXlxubm6qWbOm1q1bJ0k6e/asevbsqVKlSsnFxUVBQUFasGBBId0JgKLg4MGDOnXJUaW7TVKxSnV0MWazTi94QS5Vm6r4Q/3l6e2jzz//XEFBQdq6dasOHz4sb29vrV27VkuWLFFycrK++eYbSdKmTZvk7u4uR0dHDRo0SPv377fOEJOkc+fOady4cdq4caO2bt2qoKAgSdLgwYPVsmVLde/eXRs3bpRhGIqIiNDrr7+uNm3aaN++fTpy5IjmzJmT6314eHhoz549+uSTT+Tn56dx48apbt26OnfuXL5+fgAAAPmJmWQA7NrNevEEBQXp+eefV3p6urZs2SI3NzfFxMTI3d1dkjR27FjFxMToyy+/VMmSJXXs2DFdunSpkO8IwL3MoZiryjzxmiSpWIVaOjnjcZmcXVSq40gZkjxqNdLBuQOVkZGhfv36ydnZWS1atNAjjzwiSQoICFCTJk3022+/qWLFinJwcJBhGIqPj1edOnVsrlW5cmXrEkxJKlmypJycnBQdHa3q1avr+++/V/369bVnzx5Vq1ZNLVq00KhRo7LV7OzsrIyMjGzbHR0d1bZtW7Vt21bjx4+Xt7e3Nm3apMcff/wufmIAAAAFh5AMgN3KSy+eEydOqEuXLtb+QJUrV7Y5PiQkRA0aNJAk+fv7F/g9ALi3pV+1aNGOeB1Puqi0KxkyeZSy7jM5OkuGSSZ3HyXvWC6zf4guu3hIklJSUuTg4KD09HStWLFChmHYnPfMmTM277dv367p06erc+fO1gb6156KGRkZqYyMDF24cEGZmZlavHixzGaz0tLS1LNnT9WoUUO9e/fWjBkzFBISoj/++EMbN25UnTp19Oijj8rf31+pqanauHGj6tatK1dXV23atEm//PKLWrZsqeLFi2vdunWyWCyqVq1aPn6aAAAA+YuQDIBduL5JdsqlK8rMzNTBgweVkZGhqlWr2oy9vhfP0KFDNWjQIH311Vdq27atunTpYp2tMWjQIHXp0kV79uzRww8/rM6dO6tp06YFfm8A7k1T18UoYmucLP//kMr0DIscTQ7ZxhkOzko7eUgpu/8ny+WLkqQ2bdpo79698vPzU4MGDRQdHS0HBwd16dJFR44cUYUKFWzOMWLECO3evVsTJ06Ug0PWNf7xj39IyuqFNm3aNO3du1cZGRn6+uuvFRkZqSeeeEKStGDBAr366qsaMWKEfvvtN5UsWVKNGzdWhw4dJElNmzbVc889p+7du+vPP//U+PHj1bZtW61atUoTJkxQWlqagoKC9Mknn6hmzZr58lkCAAAUBCMzL88XL0JSUlLk5eWl5ORkeXp6FnY5AO4BUYcSNHFNjBKS0yRJpz8ereIVqqrvY+30xughOnz4sPWXymvc3d2tT3I7efKk1q5dq6+++kpffPGFZsyYoRdeeEGS9Mcff2jdunXasGGDVq5cqeeff15vvvlmwd4ggHvO1HUxem9LnM22E7O7ydHLV2Wffsu67fj0f8glsJFKP/6fv7a93kGjR4/Wli1btHv3biUnJ8tsNud4nRIlSqhcuXI6cOCAddvs2bP14osvKj4+3vq0TEkKDQ3V999/b10W7uTkpAcffFDbtm27K/cMAABwr8prVkTjfgD3tahDCRq0eI81ILvmwuWrWvSTSRkZGUpMTFRgYKDN61pAJkkVKlTQc889p1WrVmnEiBGKiIiw7itVqpT69OmjxYsXa/bs2Xr//fcL7N4A3JvSr1oUsTXu1gNvYEjy8/orDJs7d66uXLmiwMBALVy4UJs2bdJrr72moKAgpaenS5JKly6tX375Rdu2bdPRo0d19erVPF+vb9++2r59u7p06aINGzZoyZIl6tq1623XDdyof//+2ZYI3ygwMFB+fn55Ot+2bdtkGIaWLVt2N8oDACBXLLcEcN/KsGRq4poY5TZd1smnnErWa3PTXjzDhw9X+/btVbVqVZ09e1abN29WjRo1JEnjxo3TAw88oJo1a+ry5cv64osvrPsA2K9FO+KtSyzz6lqcML5jsNr/O+vr+vXra/369erTp4+efvppZWZmytnZWTVr1pSjY9aPcLNnz1bXrl3VokULSbL2I8uLiIgIXbx4UZ9++qlWrVolk8mkWrVq3V7hwB3atGmT7rMFLQCA+wAhGYD71s64pGwzyK6XKcm17QtqnbYt1148GRkZev755/Xrr7/K09NT4eHhmjVrlqSsJ76NGTNG8fHxcnFxUYsWLbR06dKCuDUA97DjSRdz3F5x+PJs2yqN+lyS5Otl1viOwQqv5WcTHLRr106nTp3K9VphYWE6f/68zbbmzZtbm/ZfLzo6Otu2JUuWaMmSJbmeH8gvFStWLOwSAADIhuWWAO5biedzDsh8n5wmn7YDJUmGg6P+0W+44uLilJ6erlOnTmnVqlXWp1nOnTtXx44dU1pamhITE/XRRx9Zm/r/5z//UUxMjC5evKg///xTq1evVkBAQMHcHIB7ViUf1zyN69W4ouY8UU+fDGisbS+3VnitvC09AwrDuHHjZBiGdanvsmXLZBiGGjdubB1TrVo1m/8ffO2111SsWDEZhqGSJUtq37591n03Lre8evWqHnnkETk5OckwDDk6Oqpdu3Y2Nfzwww8qXry4DMOQi4sLLQ4AAHcdIRmA+1Zpj5wbXd/pOADIi15N/GW6eTsmmQxpbIea6lSvnJpUKSGHWx0AFLL+/ftLknXG9LWQ7NChQ9YxcXFxatmypfX9zJkzFRERocWLF+v8+fPq3Llzrudv1qyZoqKi1LdvX23cuFFvvfWWypYtazNmzpw5GjZsmKKiolSiRAkNHjxYaWm5zxgHAOB2EZIBuG81CvCRn5dZuf3qea1JdqMAn4IsC8B9ztnRpAEtbj6rdECLADk78mMY7m3pVy36YOsvGve/Q9pw/KpcXFy0fHnWsuHvvvtObdu21YULF3T69Gnt3r1bV65c0VNPPWU9fsWKFerdu7d69uypxx57TCdPnszxOqdOndLOnTvVq1cvRUREqHXr1ho8eLAWLlxoM65Hjx6aMGGCwsLC9M477ygjI0ObNm3Kvw8AAGB3+OkMwH3LwWRofMdgScoWlF3fJJsZHID06aefqnbt2nJxcVGJEiWsv/xaLBZNmjRJ5cuXV7FixVSvXj1FRUVZj4uPj5dhGFq+fLlatGghFxcXNWzYUD/99JN27dqlBg0ayN3dXe3bt9cff/xhc8358+erRo0aMpvNql69ut55552Cvu18M+aRYD3bMiDbjDKTIT3bMkBjHgkunMLuAm9vb4WEhBR2GchnU9fFqPrYLzV57RF9tOO4Jq89ogzPsoretkMWi0WnT5/W8OHDZTabNX/+fC1YsEAmk8lmiWTr1q2tX1eqVEkWiyXHa61fv16S1K9fv5vW9NBDD1m/vtYWITY29o7vEQCAG9G4H8B9LbyWn959qr4mromxaeJ/fZNswN4lJCSoR48emj59uh577DGdP39eW7duVWZmpubMmaMZM2bovffeU0hIiD788EP94x//0OHDhxUUFGQ9x/jx4zV79mxVrFhRzzzzjJ588kl5eHhozpw5cnV1Vbdu3TRu3Di9++67krIaxo8bN07z5s1TSEiI9u7dqwEDBsjNzU19+vQprI/irhrzSLBGPFxdi3bE63jSRVXycVWvJv5FfgbZDz/8IBcXF0mSo6OjOnbsqM8++6yQq8LdNHVdjN7bEpdtuzngAaV8t1xPjpkpwzD0yCOPqHr16vriiy+UnJysChUq5HpOkyn3f+89PT3zVJfZ/Fd7hGvny8jIyNOxAADkBSEZgPteeC0/tQv21c64JCWeT1Npj6wllswgA7IkJCTo6tWrevzxx1WpUiVJf83SePPNN/Xyyy/riSeekCS9/vrr2rx5s2bPnq23337beo6RI0cqLCxMkjRs2DD16NFDGzduVLNmzSRlzRCJjIy0jh8/frxmzJihxx9/XJIUEBCgmJgYvffee/dNSCZlLb3s16JyYZdxV1WpUqWwS0A+Sr9qUcTW7AGZJLnXfVgp3y3XyvdnqFy5cpKk8PBwzZ07V+np6erWrdsdXbNNmzaSpA8++MCmpxkAAAWtaP8pEwDyyMFkqEmVEjTJLqJ8fX1lGPwzu9syLJnaEfun4jNLqkHTlqpdu7b++c9/KiIiQmfPnlVKSopOnTplDbquadasmY4cOWKzrU6dOtavy5QpI+mvoO3atsTEREnShQsXFBsbq379+snd3d36evXVV1k6VQRcW27p7e2tjIwMrV69WoZhWP8b3b59u8qUKSOTySTDMGQ2mzVx4sRCrhp5tWhHvCyZOe9z8vaVHJ119dxp+VWrJymrof+FCxd05coV9ezZ846u6e3trdDQUC1atEgDBw7Upk2bNH/+fPXt2/fObgIAgDvETDIAAOxQ1KEEm2XImc3/pYq1Y+VsHNfcuXP1yiuvaMOGDXk+n5OTk/Xra2HJjduu9SNKTU2VJEVEROjBBx+0OY+Dg8Od3RAK3A8//KBq1aqpVatWmjVrlnX7448/rqtXr2rFihUqUaKENm3aJB8fHpBSVBxPunjT/U4+5XUl8RdVbhwuKWtmodlsVnp6utq3b3/H192wYYPat2+vDz/8UBEREXJwcLDOMAMAoKAQkgEAYGeiDiVo0OI9un6yiGEYuuAdqG8VqHmLXtKzjzbWxo0bVbZsWW3fvl2tWrWyjt2+fbsaNWp0x9cvU6aMypYtq19++eWOZ56g4KRftdj0Vbvm2rJLT09Pm5mEKSkpaty4sbp06SJJCg0NLdB68fdc/884J2WffkuS1PbhGtZtly5dshkzf/58zZ8/32bb1KlTNXXqVOv7Y8eO2ex3dHTMNZhv3ry5MjNtp7dVqlQp2zYAAP4ullsCAO6qkJAQ69Kra68xY8ZIkiZNmiTDMBQYGGiz/1ovq2uuLa+89rpw4UJh3Mp9KcOSqYlrYmwCssunjip5x3KlJfysqymJevH19/XHH3+oRo0a+te//qXXX39dy5Yt09GjRzV69Gjt27dPw4YN+1t1TJw4UVOnTtVbb72ln376SQcPHtSCBQs0c+bMv3eDuKtyesJhStoV/Z6Slusx3bp1U3R0tDw9PdWqVSt9+umnBVgx/q5eTfyzPZX1RiYjaxwAAPcbQjIAwF11/vx5ValSRaNGjVKPHj0kSdOmTdPFi38t4YmNjVXNmjU1fPhwSdJXX32lM2fOSJIeffRR/f7773J1ddXw4cPl6elpXZ6Hv29nXJLNk14lyeTsqrSTh5T46QT9+v6zilv/oYaMnqT27dtr6NCheumllzRixAjVrl1bUVFR+vzzz22ebHkn+vfvr/nz52vBggWqXbu2WrVqpcjISAUEBPyt8+LuufaEw5z6U51JTdfUdTE5Hrdw4UJ999136tChg37++Wf985//VNeuXfO5Wtwtzo4mDWhx8/8OB7QIKPJPaQUAICdG5n02TzklJUVeXl5KTk7O8+OkAQB3LvniFT0TuVOnktNU1susD/s2kpfrX72oli1bpieeeEJPP/20/P39NX78eJUqVcraxH3gwIGKiIjQ888/r3nz5slkMikzM9NmGU1O23Bn/rfvNw1buu+W4+Y8UU+d6pXL/4JwT0q/alH1sV/mGJCdmN1Njl6+Kv/MW/p1xmMKCwvTmjVrcj1XkyZNtG/fvmxL8nBvm7ouRhFbbUNSk5EVkI15JLjwCgMA4A7kNSuiJxkA4I61emOTjv/51y++CclpKh/aTam7Vmcbu2vXLvn7+0vK+qX5mk6dOikiIkIHDx6UJGVmZspsNtsc6+HhoZSUlLt/A3aotIf51oNuYxzuTzd7wuE1lkzJ2cVNu3fv1u7du+Xh4aFq1aopJCREXbt2VYsWLXTy5EkdPHhQpUqVKpjCcdeMeSRYIx6ubtOPrlcTf2aQAQDua4RkAIA7cmNAds21gMy9fHU91SFUV69e1fz585Wenm4d4+bmZv3aZMr6hSsjIyN/C4YkqVGAj/y8zDqdnKacMhBDkq+XWY0CeBqhPbvVEw6vCX1quDZ+OE0NGzaUlBVyZ2RkaMKECbp69aoMw1BAQICioqLys1zkE2dHk/q1qFzYZQAAUGD4UxAA4LYlX7ySc0D24zZJknP5WirR801Nm/GWYmNjb+vchmEoLc22Z9b58+fvvFjYcDAZGt8xa6nUjb25r70f3zFYDrfq3I372s2ecFhx+HLrEw4fe6KXLl26ZLMc+sCBA7py5YoyMzNlsVgUGxv7t3vYAQAAFARCMgDAbXsmcmeO24tVqS9JSv/1kP7cvEABNUO0efPm2zr3tSdduru7a+TIkfL29qYX2V0WXstP7z5VX75etksqfb3Mevep+gqv5VdIleFewRMOAQCAPWK5JQDgtp264emI1zg5ucrJN0hXTv+s1J0rJUktWrTQ1q1b83zuL7/8UmXKlFFiYqJmzJghKSsw4wmXd1d4LT+1C/bVzrgkJZ5PU2mPrCWWzCCD9NcTDt/bEpfrGJ5wCAAA7jc83RIAcNu6vLNdP5w4d8txD1T01srBzfK/IAD5giccAgCA+0FesyJCMgDAbUu+eEV1J311y3H7xz0sL1enAqgIQH5Jv2rhCYcAAKBIy2tWxHJLAMBt83J1UqUSLjk277+mUgkXAjLgPsATDgEAgL3gz4AAgDvyzb9aq1IJlxz3VSrhom/+1bqAKwLsS2hoqIYPH57n8fHx8TIMQ/v27cu3mgAAAIoyllsCAP6W5ItX9EzkTp1KTlNZL7M+7NuIGWTAbfD399fw4cNvK/CSpKSkJDk5OcnDwyNP4zMyMvTHH3+oZMmScnRkMQEAALAfLLcEABQIL1cnmvMDhcDHx+e2xjs4OMjX1zefqgEAACj6WG4JAABwExaLRdOnT1dgYKCKFSumihUr6rXXXpMkHTx4UK1bt5aLi4tKlCihgQMHKjU11Xps37591blzZ7355pvy8/NTiRIl9Pzzz+vKlSuSspZMHj9+XC+++KIMw5BhGJKkP//8Uz169FC5cuXk6uqq2rVr65NPPrGp68bllv7+/poyZYqeeeYZeXh4qGLFinr//fet+29cbhkdHS3DMLRx40Y1aNBArq6uatq0qY4ePWpznVdffVWlS5eWh4eH+vfvr9GjR6tevXp36+MFAAC4ZxCSAQAA3MSYMWM0bdo0jR07VjExMfr4449VpkwZXbhwQWFhYSpevLh27dqlFStW6Ouvv9aQIUNsjt+8ebNiY2O1efNmLVy4UJGRkYqMjJQkrVq1SuXLl9ekSZOUkJCghIQESVJaWpoeeOABrV27VocOHdLAgQPVq1cv7dy586a1zpgxQw0aNNDevXs1ePBgDRo0KFvodaNXXnlFM2bM0O7du+Xo6KhnnnnGum/JkiV67bXX9Prrr+uHH35QxYoV9e67797BpwgAAHDvoycZAADAdTIsmdoZl6TE82lyM66ow4PVNW/ePPXv399mXEREhF5++WWdPHlSbm5ukqR169apY8eOOnXqlMqUKaO+ffsqOjpasbGxcnBwkCR169ZNJpNJS5culZT3nmQdOnRQ9erV9eabb0rKmklWr149zZ4923qeFi1aaNGiRZKkzMxM+fr6auLEiXruuecUHx+vgIAA7d27V/Xq1VN0dLQeeughff3112rTpo21/kcffVSXLl2S2WxW48aN1aBBA82bN89aR/PmzZWamsoDAAAAQJFBTzIAAIDbFHUoQRPXxCghOU2SdPnUUV2+fFmO5WtnG3vkyBHVrVvXGpBJUrNmzWSxWHT06FGVKVNGklSzZk1rQCZJfn5+Onjw4E3ryMjI0JQpU7R8+XL99ttvSk9P1+XLl+Xq6nrT4+rUqWP92jAM+fr6KjExMc/H+Pn5SZISExNVsWJFHT16VIMHD7YZ36hRI23atOmm5wQAACiKCMkAAACUFZANWrxH10+xN5yKSZL+s/qQfMtXVHgtv9s+r5OT7dNeDcOQxWK56TFvvPGG5syZo9mzZ6t27dpyc3PT8OHDlZ6eftevdf0x13qi3eoYAACA+xE9yQAAgN3LsGRq4poY3diDwql4WRmOxZR2fL8mrolRhuWvETVq1ND+/ft14cIF67bt27fLZDKpWrVqeb62s7OzMjIybLZt375dnTp10lNPPaW6deuqcuXK+umnn+7o3v6OatWqadeuXTbbbnwPAABwvyAkAwAAdm9nXJJ1ieX1DEdneT7YRWejF+jn7Wu1KvoHfffdd/rggw/Us2dPmc1m9enTR4cOHdLmzZv1wgsvqFevXtallnnh7++vLVu26LffftOZM2ckSUFBQdqwYYO+/fZbHTlyRM8++6x+//33u3a/efXCCy/ogw8+0MKFC/Xzzz/r1Vdf1YEDB6wzzgAAAO4nhGQAAMDuJZ7PHpBd49XsCXk2fEznti7Rk2FN1b17dyUmJsrV1VXr169XUlKSGjZsqK5du6pNmzY2Te7zYtKkSYqPj1eVKlVUqlQpSdJ//vMf1a9fX2FhYQoNDZWvr686d+78d27xjvTs2VNjxozRyJEjVb9+fcXFxalv374ym80FXgsAAEB+4+mWAADA7u2I/VM9Ir675bhPBjRWkyolCqCie1e7du3k6+trfYomAADAvY6nWwIAAORRowAf+XmZdTo5LVtfMkkyJPl6mdUowKegSytUFy9e1H//+1+FhYXJwcFBn3zyib7++mtt2LChsEsDAAC461huCQAA7J6DydD4jsGSsgKx6117P75jsBxM9tWLyzAMrVu3Ti1bttQDDzygNWvWaOXKlWrbtm1hlwYAAHDXsdwSAADg/0UdStDENTE2Tfz9vMwa3zFY4bX8CrEyAAAA3CmWWwIAANym8Fp+ahfsq51xSUo8n6bSHllLLO1tBhkAAIA9IiQDAAC4joPJsPvm/AAAAPaInmQAAAAAAACwe4RkAAAAAAAAsHuEZAAAAAAKRVRUlJo3by5vb2+VKFFCHTp0UGxsrCQpPT1dQ4YMkZ+fn8xmsypVqqSpU6cWcsUAgPsZIRkAAACAQnHhwgW99NJL2r17tzZu3CiTyaTHHntMFotFb731lj7//HMtX75cR48e1ZIlS+Tv71/YJQMA7mM07gcAAABQIDIsmTZPj+382OM2T4/98MMPVapUKcXExOjEiRMKCgpS8+bNZRiGKlWqVIiVAwDsASEZAAAAgHwXdShBE9fEKCE5zbrN+8oZeR/5TCd+PKAzZ87IYrFIkk6cOKG+ffuqXbt2qlatmsLDw9WhQwc9/PDDhVU+AMAOsNwSAAAAQL6KOpSgQYv32ARkknQ48j/6/sgJDfj3NH3//ff6/vvvJWX1I6tfv77i4uI0efJkXbp0Sd26dVPXrl0Lo3wAgJ0gJAMAAACQbzIsmZq4JkaZN26/lKKrSb/Ku2l3ff5HCVWtVl1nz561GePp6anu3bsrIiJCy5Yt08qVK5WUlFRwxQMA7ArLLQEAAADkm51xSdlmkEmSyewuk4unzu9frxPuPnrn4xQteuuvp1fOnDlTfn5+CgkJkclk0ooVK+Tr6ytvb+8CrB4AYE+YSQYA/y8yMpIfvAEAuMsSz2cPyCTJMEwq+Y9RSj99TKc+eF4zJ72iN954w7rfw8ND06dPV4MGDdSwYUPFx8dr3bp1Mpn4FQYAkD+MzMzMG2c+F2kpKSny8vJScnKyPD09C7scAEXIpUuXdP78eZUuXTrPx4SGhqpevXqaPXt2/hUGAEARtiP2T/WI+O6W4z4Z0FhNqpQogIoAAPYmr1kRf4YBgP/n4uJyWwEZAAC4tUYBPvLzMsvIZb8hyc/LrEYBPgVZFgAA2RCSAbhvhIaGasiQIRoyZIi8vLxUsmRJjR07VtcmzJ49e1a9e/dW8eLF5erqqvbt2+vnn3+2Hn/jcssJEyaoXr16WrRokfz9/eXl5aUnnnhC58+flyT17dtX33zzjebMmSPDMGQYhuLj4wvylgEAuOc5mAyN7xgsSdmCsmvvx3cMloMptxgNAICCQUgG4L6ycOFCOTo6aufOnZozZ45mzpyp+fPnS8oKtXbv3q3PP/9cO3bsUGZmph555BFduXIl1/PFxsZq9erV+uKLL/TFF1/om2++0bRp0yRJc+bMUZMmTTRgwAAlJCQoISFBFSpUKJD7BArD7NmzZRiGjh8/XtilAChiwmv56d2n6svXy2yz3dfLrHefqq/wWn6FVBkAAH/h6ZYAirQMS6Z2xiUp8XyaUi5dUYUKFTRr1iwZhqFq1arp4MGDmjVrlkJDQ/X5559r+/btatq0qSRpyZIlqlChglavXq1//vOfOZ7fYrEoMjJSHh4ekqRevXpp48aNeu211+Tl5SVnZ2e5urrK19e3wO4ZAICiKLyWn9oF+1r/f7u0R9YSS2aQAQDuFYRkAIqsqEMJmrgmxvpY+dMJKfIsXVHrD5+2/kW6SZMmmjFjhmJiYuTo6KgHH3zQenyJEiVUrVo1HTlyJNdr+Pv7WwMySfLz81NiYmI+3RFgn1JTU+Xu7l7YZQAoAA4mg+b8AIB7FsstARRJUYcSNGjxHmtAds2l9AwNWrxHUYcS7sp1nJycbN4bhiGLxXJXzg3klbe3t+rUqaOQkBCZTCaZTCb17t1biYmJCgoKkmEYcnJy0qRJk6zHfPbZZypVqpQMw5DJZFLlypV19OjRv3XOaxYsWCAXFxcZhiF3d3d99tlnNvvfeecdeXl5yTAMOTo6qm7dujbhsqOjo9q0aaMqVarIMAzVq1dPqampqlOnjhwcHKzHhYWF5cOnCQAAAOSMkAxAkZNhydTENTHKzGHf5VM/SZImrolRhiVT3333nYKCghQcHKyrV6/q+++/t479888/dfToUQUHB99xLc7OzsrIyLjj44G8OnjwoEqUKKH169erdevWWrRokWrVqqUmTZroq6++UrVq1TRhwgSdOXNGx48fV5cuXRQUFKS1a9dqyZIlSk5OVuPGje/4nNebOnWqxo8fr08//VRubm7q1q2bLl68KEnatGmTnn/+ebVu3VpfffWV3n77bf38889q1qyZzTk2bdqk4OBgbdy4Ue+//76eeOIJxcTEaM6cOdq2bZveeustVa1aNX8/VAAAAOA6LLcEUOTsjEvKNoPsmqvn/9CfGyOUXq+9Xn0rVnPnztWMGTMUFBSkTp06acCAAXrvvffk4eGh0aNHq1y5curUqdMd1+Lv76/vv/9e8fHxcnd3l4+Pj0wm/v6Avy/9qkWLdsTreNJFpV+1yNPTU19//bUkqVWrVipWrJg8PDz00UcfSZI+/vhj1a1bV6tXr9bSpUvl4+Ojb7/91nq+wMBANWrUSOvXr7fO0Lqdc/bv3996rueee06jR4+WJNWrV0+BgYF65ZVXNGvWLA0aNEjVq1e3mV2WkZGh559/XufOnbM+QdbPz09r1qyxjnnxxRfl6empwYMHy2QyZQvVAAAAgPzGb3IAipzE8zkHZJLkVrO1Mq+mK+Gjl/Tm+FEaNmyYBg4cKClridgDDzygDh06qEmTJsrMzNS6deuyLam8HSNHjpSDg4OCg4NVqlQpnThx4o7PBVwzdV2Mqo/9UpPXHtFHO44r7WqGLjkX19R1MZKyZjA6ODioevXq1mNq1aolSYqLi9OPP/6oP//8U4ZhWF+NGjWSJJvZlBUrVrR+fatzXu+JJ56wfl2lShW5uLho3759kqRff/1VP/74o821n3/+eUnStm3brMfVrFnT5pyjRo3SuXPnZDabVbduXU2dOvU2PzUAAADg72EmGYAip7SHOdd9hslBPm0HqUTY8/pkQGOb5sDFixe3zpDJSd++fdW3b1/r+wkTJmjChAk2Y4YPH67hw4db31etWlU7duy47XsAcjN1XYze2xKXfYfJwbp9zCNZS4SdnZ3/2v3/MxgtFosuX76ssmXL5vjve+3ata1fOzpm/zEgt3Pm1dWrV1WnTh3NnDkz277rH5xxY6P+nj17qlWrVnrjjTe0fv16/fvf/9bbb7+tX3/9Nc/XBgAAAP4OQjIARU6jAB/5eZl1Ojktx75khiRfr6zHygNFSfpViyK25hCQXSdia5xGPFz9pmOqVq2q3bt3q1mzZjKbcw+V79Ty5cvVpEkTSVmzzC5duqR69epJylpGefLkSbVp0+a2z1u+fHnNmTNHkvTaa6/pP//5j2JjY1WlSpW7VjsAAACQG0IyAEWOg8nQ+I7BGrR4jwwpx6BsfMdgOZiMgi4N+FsW7YiXJad/oa9jycwadzNz585VgwYNFBgYqNdee00VKlTQjh07FBkZqcOHD9vMFrsT7777rsqWLavAwEANHjxYDg4Omjx5siTpzTff1D//+U/VqVNHr7zyinx8fLR582Z98cUXOnDgQK7n7NSpkypWrKj27dvLwcFBH330kUwmkypVqvS3agUAAADyipAMQJEUXstP7z5VXxPXxFib+Ps+OU1+XmaN7xis8Fp+hVwhcPuOJ128K+Pq16+v9evXq0+fPnr66aeVmZkpZ2dn1axZM8cllrdr1KhRGjt2rC5fviw3NzctXbrUunyya9euioyM1EsvvWTtXWY2m9WiRYubntPLy0sffPCB5s2bJylrefTixYvvSr0AAABAXhiZmZm3+Jt10ZKSkiIvLy8lJyfL09OzsMsBkM8yLJnaGZekxPNpKu2RtcSSGWQoqj7Y+osmrz1yy3FjH62hfi0qF0BFAAAAQNGX16yIP88CKNIcTIZNc36gKOvVxF+vrTty0yWXJiNrHAAAAIC7y1TYBQAAgCzOjiYNaBFw0zEDWgTI2ZH/+wYAAADuNn7KBgDgHjLmkWA92zJAN64aNhnSsy0DNOaR4MIpDLjPREdHyzAMnTt3rrBLAQAA9wh6kgEAcA9Kv2rRoh3xOp50UZV8XNWriT8zyIA7FBoaqnr16mn27NnWbenp6UpKSlKZMmVkGPSyBADgfkZPMgAAijBnRxPN+XFfSE9Pl7Oz8z13bWdnZ/n6+hZwRQAA4F7Gn6QBAACQZ6GhoRoyZIiGDBkiLy8vlSxZUmPHjtW1xQn+/v6aPHmyevfuLU9PTw0cOFCStHLlStWsWVPFihWTv7+/ZsyYYXPea8f16NFDbm5uKleunN5++22bMefOnVP//v1VqlQpeXp6qnXr1tq/f791/4QJE1SvXj3Nnz9fAQEBMpvN6tu3r7755hvNmTNHhmHIMAzFx8fnuNxy27ZtatGihVxcXFShQgUNHTpUFy5csO5/5513FBQUJLPZrDJlyqhr1653++MFAACFiJAMAAAAt2XhwoVydHTUzp07NWfOHM2cOVPz58+37n/zzTdVt25d7d27V2PHjtUPP/ygbt266YknntDBgwc1YcIEjR07VpGRkTbnfeONN6zHjR49WsOGDdOGDRus+//5z38qMTFRX375pX744QfVr19fbdq0UVJSknXMsWPHtHLlSq1atUr79u3TnDlz1KRJEw0YMEAJCQlKSEhQhQoVst1TbGyswsPD1aVLFx04cEDLli3Ttm3bNGTIEEnS7t27NXToUE2aNElHjx5VVFSUWrZseZc/WQAAUJjoSQYAAICbyrBkamdckhLPp2nis92Udv6sDh8+bO3lNXr0aH3++eeKiYmRv7+/QkJC9Nlnn1mP79mzp/744w999dVX1m2jRo3S2rVrdfjwYUlZM8lq1KihL7/80jrmiSeeUEpKitatW6dt27bp0UcfVWJioooVK2YdExgYqFGjRmngwIGaMGGCpkyZot9++02lSpWyjsmpJ1l0dLQeeughnT17Vt7e3urfv78cHBz03nvvWcds27ZNrVq10oULF7Ru3To9/fTT+vXXX+Xh4XH3PlwAAJDv8poVMZMMAIACkp6eXtglALct6lCCmr++ST0ivtOwpfsUk5CiM64Vtf7waeuYJk2a6Oeff1ZGRoYkqUGDBjbnOHLkiJo1a2azrVmzZjbHXDvP9Zo0aaIjR45Ikvbv36/U1FSVKFFC7u7u1ldcXJxiY2Otx1SqVMkmIMur/fv3KzIy0ubcYWFhslgsiouLU7t27VSpUiVVrlxZvXr10pIlS3Tx4sXbvg4AALh3FUhI9vbbb8vf319ms1kPPvigdu7cmevYyMhIa7+Iay+z2VwQZQIA7EhoaKiGDh2qUaNGycfHR76+vpowYYJ1/616H8XGxqpTp04qU6aM3N3d1bBhQ3399dc218itNxNQVEQdStCgxXuUkJxms/1SeoYGLd6jqEMJOR7n5uZ212tJTU2Vn5+f9u3bZ/M6evSo/vWvf/3ta6empurZZ5+1Off+/fv1888/q0qVKvLw8NCePXv0ySefyM/PT+PGjVPdunVtepoBAICiLd9DsmXLlumll17S+PHjtWfPHtWtW1dhYWFKTEzM9RhPT09rz4iEhAQdP348v8sEANihhQsXys3NTd9//72mT5+uSZMmWfsf3ar3UWpqqh555BFt3LhRe/fuVXh4uDp27KgTJ07YXOPG3kxAUZFhydTENTHKqS/H5VM/SZImrolRhiVT3333nYKCguTg4JDjuWrUqKHt27fbbNu+fbuqVq1qc8x3331nM+a7775TjRo1JEn169fX6dOn5ejoqMDAQJtXyZIlb3ovzs7ONjPWclK/fn3FxMRkO3dgYKD1CZmOjo5q27atpk+frgMHDig+Pl6bNm266XkBAEDR4ZjfF5g5c6YGDBigp59+WpL03//+V2vXrtWHH36o0aNH53iMYRg8khsAcFdd31OptIdZmZLq1Kmj8ePHS5KCgoI0b948bdy4US4uLtq5c6dN76M333xTq1ev1qeffqqBAweqbt26qlu3rvX8kydP1meffabPP//c2uhbklq3bq0RI0YU6L0Cd8POuKRsM8iuuXr+D/25MULp9drr1bdiNXfu3GxPq7zeiBEj1LBhQ02ePFndu3fXjh07NG/ePL3zzjs247Zv367p06erc+fO2rBhg1asWKG1a9dKktq2basmTZqoc+fOmj59uqpWrapTp05p7dq1euyxx7It8byev7+/vv/+e8XHx8vd3V0+Pj7Zxrz88stq3LixhgwZov79+8vNzU0xMTHasGGD5s2bpy+++EK//PKLWrZsqeLFi2vdunWyWCyqVq1aXj5OAABQBORrSJaenq4ffvhBY8aMsW4zmUxq27atduzYketxqampqlSpkiwWi+rXr68pU6aoZs2aOY69fPmyLl++bH2fkpJy924AAHBfiDqUoIlrYmx+4U86cVatGoXYjPPz81NiYqJN76PrXbp0ydr7KDU1VRMmTNDatWuVkJCgq1ev6tKlS9lmkt3sF3fgXpZ4PueATJLcarZW5tV0JXz0kt40O2nYsGE3XU5cv359LV++XOPGjdPkyZPl5+enSZMmqW/fvjbjRowYod27d2vixIny9PTUzJkzFRYWJinrj6jr1q3TK6+8oqefflp//PGHfH191bJlS5UpU+am9zJy5Ej16dNHwcHBunTpkuLi4rKNqVOnjr755hu98soratGihTIzM1WlShV1795dkuTt7a1Vq1ZpwoQJSktLU1BQkD755JNcf0YFAABFT76GZGfOnFFGRka2H1zKlCmjH3/8McdjqlWrpg8//FB16tRRcnKy3nzzTTVt2lSHDx9W+fLls42fOnWqJk6cmC/1AwCKvms9lW5cMpZ+1aJvjp1V1KEEhdfyk5T1S7jFYrH2PoqOjs52Pm9vb0lZv3Rv2LBBb775pgIDA+Xi4qKuXbtma86fH72ZgIJQ2iP3nrCGyUE+bQepRNjz+mRAYzWp8legHB8fn+MxXbp0UZcuXW56TU9PTy1fvjzX/R4eHnrrrbf01ltv5bh/woQJNr0Fr6latWq2P9D6+/vrxoe8N2zY0OYJnNdr3rx5jt8TAADA/SPfl1veriZNmtg82ahp06aqUaOG3nvvPU2ePDnb+DFjxuill16yvk9JSVGFChUKpFYAwL3tZj2Vrpm4Jkbtgn3lYDKs267vfeTv75/jcdu3b1ffvn312GOPScqaWZZbOAAURY0CfOTnZdbp5LQc/xsyJPl6mdUoIPvSRQAAgKIoXxv3lyxZUg4ODvr9999ttv/+++957jnm5OSkkJAQHTt2LMf9xYoVk6enp80LAADp5j2VrklITtPOuCSbbdf3Pvrqq68UHx+vb7/9Vq+88op2794tKauH2apVq6xPwHvyySdlsVjy7V6AguZgMjS+Y7CkrEAsJ+M7BtsEzAAAAEVZvoZkzs7OeuCBB7Rx40brNovFoo0bN9rMFruZjIwMHTx4UH5+fvlVJgDgPnWznko3G3et91HLli319NNPq2rVqnriiSd0/PhxawuBmTNnqnjx4mratKk6duyosLAw1a9f/67fA1CYwmv56d2n6svX66+ll75PTlPNLkP17lP1rUuV74b4+HgNHz78rp0PAADgdhmZNzZjuMuWLVumPn366L333lOjRo00e/ZsLV++XD/++KPKlCmj3r17q1y5cpo6daokadKkSWrcuLECAwN17tw5vfHGG1q9erV++OEHBQcH3/J6KSkp8vLyUnJyMrPKAMDO7Yj9Uz0ivrvluBt7KgGwdePTYRsF+DCDDAAAFBl5zYryvSdZ9+7d9ccff2jcuHE6ffq06tWrp6ioKOtf4k+cOCGT6a8JbWfPntWAAQN0+vRpFS9eXA888IC+/fbbPAVkAABcj55KwN3hYDIIkgEAwH0v32eSFTRmkgEArnft6ZaSbIKya3Ng7vaSMQAAAAD3lrxmRfnakwwAgMKWU08lKWsGGQHZ/S80NNSu+lz17dtXnTt3LuwyAAAAiiRmkgEA7AI9lexTUlKSnJyc5OHhccux8fHxCggI0N69e1WvXr38L+5vyK3W5ORkZWZmytvbu9BqAwAAuNfcMz3JACA/hIaGql69epo9e3Zhl4Iigp5K9snHp3D6zV25ckVOTk4Ffl0vL68CvyYAAMD9guWWAADgvnX9ckt/f39NmTJFzzzzjDw8PFSxYkW9//771rEBAQGSpJCQEBmGodDQUOu++fPnq0aNGjKbzapevbreeecd6774+HgZhqFly5apVatWMpvNWrJkiY4fP66OHTuqePHicnNzU82aNbVu3TrrcYcOHVL79u3l7u6uMmXKqFevXjpz5ox1v8Vi0fTp0xUYGKhixYqpYsWKeu21125a643LLS9fvqyhQ4eqdOnSMpvNat68uXbt2mXdHx0dLcMwtHHjRjVo0ECurq5q2rSpjh49+vc+eAAAgCKIkAwA7lHp6emFXQJw35kxY4YaNGigvXv3avDgwRo0aJA1ENq5c6ck6euvv1ZCQoJWrVolSVqyZInGjRun1157TUeOHNGUKVM0duxYLVy40Obco0eP1rBhw3TkyBGFhYXp+eef1+XLl7VlyxYdPHhQr7/+utzd3SVJ586dU+vWrRUSEqLdu3crKipKv//+u7p162Y935gxYzRt2jSNHTtWMTEx+vjjj61PB8+t1huNGjVKK1eu1MKFC7Vnzx4FBgYqLCxMSUlJNuNeeeUVzZgxQ7t375ajo6OeeeaZv/tRAwAAFDmEZACKLIvFolGjRsnHx0e+vr6aMGGCdd+JEyfUqVMnubu7y9PTU926ddPvv/8uKatnj4ODg3bv3m09j4+Pjxo3bmw9fvHixapQoYL1/cmTJ9WtWzd5e3vLx8dHnTp1Unx8vCTpq6++ktls1rlz52zqGzZsmFq3bm19v23bNrVo0UIuLi6qUKGChg4dqgsXLlj3+/v7a/Lkyerdu7c8PT01cODAu/VRAXYlw5KpHbF/6n/7flPKpSu6vv3qI488osGDByswMFAvv/yySpYsqc2bN0uSSpUqJUkqUaKEfH19rUs1x48frxkzZujxxx9XQECAHn/8cb344ot67733bK47fPhw6xg/Pz+dOHFCzZo1U+3atVW5cmV16NBBLVu2lCTNmzdPISEhmjJliqpXr66QkBB9+OGH2rx5s3766SedP39ec+bM0fTp09WnTx9VqVJFzZs3V//+/W9a6/UuXLigd999V2+88Ybat2+v4OBgRUREyMXFRR988IHN2Ndee02tWrVScHCwRo8erW+//VZpaWl34x8HAABAkUFIBqDIWrhwodzc3PT9999r+vTpmjRpkjZs2CCLxaJOnTopKSlJ33zzjTZs2KBffvlF3bt3l5TVs6devXqKjo6WJB08eFCGYWjv3r1KTU2VJH3zzTdq1aqVpKzeQmFhYfLw8NDWrVu1fft2ubu7Kzw8XOnp6WrTpo28vb21cuVKa20ZGRlatmyZevbsKUmKjY1VeHi4unTpogMHDmjZsmXatm2bhgwZYnNPb775purWrau9e/dq7Nix+f0RAvedqEMJav76JvWI+E7Dlu5TTEKKlu/+VVGHEiRJderUsY41DEO+vr5KTEzM9XwXLlxQbGys+vXrJ3d3d+vr1VdfVWxsrM3YBg0a2LwfOnSoXn31VTVr1kzjx4/XgQMHrPv279+vzZs325yzevXqkrK+Xxw5ckSXL19WmzZt7viziI2N1ZUrV9SsWTPrNicnJzVq1EhHjhyxGXv95+Lnl/XE15t9LgAAAPcjGvcDKBJufDJhprJ+qRs/frwkKSgoSPPmzdPGjRslZQVfcXFx1tlgH330kWrWrKldu3apYcOGCg0NVXR0tEaOHKno6Gi1a9dOP/74o7Zt26bw8HBFR0dr1KhRkqRly5bJYrFo/vz5MoyspyEuWLBA3t7eio6O1sMPP6wnnnhCH3/8sfr16ydJ2rhxo86dO6cuXbpIkqZOnaqePXtaeyMFBQXprbfeUqtWrfTuu+/KbDZLklq3bq0RI0YUyGcK3G+iDiVo0OI9uvGx3RcuX9WgxXt06UpGtmb6hmHIYrHkes5rwXlERIQefPBBm30ODg42793c3Gze9+/fX2FhYVq7dq2++uorTZ06VTNmzNALL7yg1NRUdezYUa+//nq2a/r5+emXX3651e3eVdd/Lte+z93scwEAALgfEZIBuOdFHUrQxDUxSkj+a+lP0omzatUoxGacn5+fEhMTdeTIEVWoUMFmuWRwcLC8vb115MgRNWzYUK1atdIHH3ygjIwMffPNN3r44Yfl6+ur6Oho1alTR8eOHbM2wt6/f7+OHTsmDw8Pm+ulpaVZZ5L07NlTjRs31qlTp1S2bFktWbJEjz76qLy9va3nOHDggJYsWWI9PjMzUxaLRXFxcapRo4ak7DNRAORNhiVTE9fEZAvIrnfu4hVZMnMf4ezsnHWujAzrtjJlyqhs2bL65ZdfrDNDb0eFChX03HPP6bnnntOYMWMUERGhF154QfXr19fKlSvl7+8vR8fsP44FBQXJxcVFGzdutC6xvFWtN6pSpYqcnZ21fft2VapUSVLWzNhdu3ZZA3sAAAD8hZAMwD0tt5kh6Vct+ubYWUUdSlB4raylQbeaEXK9li1b6vz589qzZ4+2bNmiKVOmyNfXV9OmTVPdunVVtmxZBQUFScqaSfLAAw/YBFzXXOsL1LBhQ1WpUkVLly7VoEGD9NlnnykyMtI6LjU1Vc8++6yGDh2a7RwVK1a0fn3jTBQAebMzLskmSL9RprKCtBN/Xsx1TOnSpeXi4qKoqCiVL19eZrNZXl5emjhxooYOHSovLy+Fh4fr8uXL2r17t86ePauXXnop1/MNHz5c7du3V9WqVXX27Flt3rzZGog///zzioiIUI8ePay9FY8dO6alS5dq/vz5MpvNevnllzVq1Cg5OzurWbNm+uOPP3T48GH169cv11qv5+bmpkGDBulf//qXfHx8VLFiRU2fPl0XL160znoFAADAXwjJANyz8jIzZOKaGLUL9pWDybBuq1Gjhk6ePKmTJ09aZ5PFxMTo3LlzCg4OliR5e3urTp06mjdvnpycnFS9enWVLl1a3bt31xdffGHtRyZJ9evX17Jly1S6dGl5enrmWkvPnj21ZMkSlS9fXiaTSY8++qjNOWJiYhQYGHiHnwaAm0k8n7cm8+cvX811n6Ojo9566y1NmjRJ48aNU4sWLRQdHa3+/fvL1dVVb7zxhv71r3/Jzc1NtWvXvuVsrIyMDD3//PP69ddf5enpqfDwcM2aNUuSVLZsWW3fvl0vv/yyHn74YV2+fFmVKlVSeHi4TKaslrFjx46Vo6Ojxo0bp1OnTsnPz0/PPffcTWu90bRp02SxWNSrVy+dP39eDRo00Pr161W8ePE8fV4AAAD2xMjMvMm6gyIoJSVFXl5eSk5OvukvswDufTti/1SPiO9y3Hf649FyLl1ZPm0H6pMBjdWkSgl17txZ3t7eWrBggerXry8PDw/Nnj1bV69e1eDBg+Xu7m7zS+SLL76ouXPnqmvXrlq6dKkkqV69ejp06JDefvttPfvss5Kkixcvql69eipXrpwmTZqk8uXL6/jx41q1apVGjRql8uXLS5KOHTumoKAg1alTRw0bNtT8+fOt1zpw4IAaN26sZ555Rv3795ebm5tiYmK0YcMGzZs3T1LW0y2HDx/OMijgDtzs+8X1rn2/AAAAgP3Ia1bE0y0B3LPyOjPkxnGGYeh///ufihcvrpYtW6pt27aqXLmyli1bZjOuVatWysjIsPYek6TQ0NBs21xdXbVlyxZVrFhRjz/+uGrUqKF+/fopLS3N5htsYGCgGjVqpAMHDmTrXVSnTh198803+umnn9SiRQuFhIRo3LhxKlu2bB4/DQA30yjAR35eZhm57Dck+XmZ1SjApyDLAgAAQBHCTDIA9yxmhgC4Hdd6GEqyWaZ9LTh796n61h6GAAAAsB/MJANQ5DEzBMDtCK/lp3efqi9fL7PNdl8vMwEZAAAAbonG/bA7EyZM0OrVq7Vv377CLgW34GAyNL5jsAYt3iNDOc8MGd8x2KZpPwD7Fl7LT+2CfbUzLkmJ59NU2iMrSOf7BAAAAG6FkAzAPe3azJCJa2KUkPxX7zFfL7PGdwxmZgiAbBxMBkuwAQAAcNtYbom7zmKxaOrUqQoICJCLi4vq1q2rTz/9VJmZmWrbtq3CwsJ0rRVeUlKSypcvr3HjxlmPX7NmjRo2bCiz2aySJUvqscces+67fPmyRo4cqXLlysnNzU0PPvigzdMKIyMj5e3trdWrVysoKEhms1lhYWE6efKkdf/EiRO1f/9+GYYhwzAUGRlZIJ8L7lx4LT9te7m1PhnQWHOeqKdPBjTWtpdbE5ABAAAAAO4aZpLhrps6daoWL16s//73vwoKCtKWLVv01FNPqVSpUlq4cKFq166tt956S8OGDdNzzz2ncuXKWUOytWvX6rHHHtMrr7yijz76SOnp6Vr3f+zdd1gU19fA8e8svSNYwIqKItgQu1iwRWzRmERjsKCiRmMUWzSxorFGbDExlgTsmsTesGMUe8EGYglKNCjGgiJSd98/eJmfK6BobOj5PA+PzMydO3eGInv23HO3bFH77tevHxEREaxcuZLChQuzdu1avL29OXPmDGXKlAEgMTGRCRMmsHjxYoyNjenbty+fffYZYWFhdOjQgbNnzxISEsLOnTsBsLGxef0PSTw3yQwRQgghhBBCCPEqyeqW4qVI1+o4En2H67fj6exVkR07dlDXs4563M/Pj8TERJYvX87vv/9Oly5d8Pf354cffuDkyZNqgKtOnTqUKlWKpUuXZrlGTEwMpUqVIiYmhsKFC6v7mzRpQo0aNZg4cSLBwcF069aNQ4cOUbNmTQDOnz+Pq6srhw8fpkaNGlKTTAghhBBCCCGEeI/kNlYkmWTiPws5G6vWi0q5dZWkR4k0aNQYYwONWig5JSWFKlWqAPDpp5+ydu1aJk+ezNy5c9UAGUB4eDg9e/bM9jpnzpwhPT2dsmXL6u1PTk7G3v5/GUaGhoZUr15d3S5Xrhy2trZERkZSo0aNl3bfQgghhBBCCCGEeHdIkEz8JyFnY+mz9IS66qAuNaOwesGPx2BgZc+4NuWpX7YgACYmJkDGdMjjx49jYGDAxYsX9fozMzPL8VoJCQkYGBio5z7O0tLyJd2REEIIIYQQQggh3kdSuF+8sHStjoCNETw+X9fIvhgYGJF6/xZG+Qqz4NQjSpYqjbOzM8WKFQNg8ODBaDQatm7dyuzZs9m9e7d6fqVKldi1a1e216tSpQrp6enExcXh7Oys9+Hg4KC2S0tL49ixY+p2VFQU9+7dw9XVFQBjY2PS09Nf4pMQQgghhBBCCCFEXieZZOKFHYm+Q2x8kt4+jYk51jXacXf3QtDpSCnqxpKNe3gQcw5ra2vy58/Pr7/+ysGDB/Hw8GDo0KF07dqV06dPky9fPsaMGUPjxo0pXbo0n332GWlpaWzZsoVhw4ZRtmxZfHx86NKlC4GBgVSpUoVbt26xa9cuKlWqRMuWLQEwMjLiq6++Yvbs2RgaGtKvXz9q1aqlTrV0cnIiOjqa8PBwihYtipWVlZrlJoQQQgghhBBCiPeTZJKJFxb3ICnb/bb1OmFTpwPxh37nn4V98O/Wns2bN+Pk5ESPHj0YO3YsHh4eAAQEBFCoUCG++OILALy8vPj999/ZsGED7u7uNGrUiCNHjqh9BwUF0aVLFwYPHoyLiwtt27bl6NGjFC9eXG1jbm7OsGHD+Pzzz/H09MTS0pJVq1apxz/++GO8vb1p2LAhBQoUYMWKFa/i8QghhBBCCCGEECIPkdUtxQs7ePk2HRccema7FT1rUbu0/TPbvQzBwcH4+/tz796913I9IYQQQgghhBBCvN1yGyuSTDLxwmqUtMPRxhQlh+MK4GhjSo2Sdq9zWEIIIYQQQgghhBDPTYJk4oUZaBTGtHYDyBIoy9we09oNA01OYTQhhBBCCCGEEEKIt4MEycR/4l3BkbmdPHCwMdXb72BjytxOHnhXcHyt4/H19ZWplkIIIYQQQgghhHhusrql+M+8KzjS1M2BI9F3iHuQREGrjCmWkkEmhBBCCCGEEEKIvEKCZOKlMNAor604vxBCCCGEEEIIIcTLJtMthRBCCCGEEEIIIcR7T4JkQgghhBBCCCGEEOK9J0EyIYQQQgghhBBCCPHekyCZEEIIIYQQQgghhHjvSZBMCCGEEO+sK1euoCgK4eHh/6kfJycnZs6c+VLGJIQQQggh3k6yuqUQQgghxDMcPXoUCwuLNz0MIYQQQgjxCkmQTAghhBDiGQoUKPCmhyCEEEIIIV4xmW4phBBCiDxPq9UydepUnJ2dMTExoXjx4kyYMEE9/tdff9GwYUPMzc2pXLkyBw8e1Dt/9erVlC9fHhMTE5ycnAgMDNQ7/uR0y3v37tG7d28KFSqEqakpFSpUYNOmTerx/fv3U69ePczMzChWrBj9+/fn4cOHr+bmhRBCCCHESyFBMiGEEELked988w2TJ09m1KhRREREsHz5cgoVKqQeHzFiBEOGDCE8PJyyZcvSsWNH0tLSADh+/Djt27fns88+48yZM4wdO5ZRo0YRHByc7bW0Wi3NmzcnLCyMpUuXEhERweTJkzEwMADg8uXLeHt78/HHH3P69GlWrVrF/v376dev3yt/DkIIIYQQ4sUpOp1O96YH8TLdv38fGxsb4uPjsba2ftPDEUIIIcQrkK7VcST6DnEPkrBQUmlVsxxz5szBz89Pr92VK1coWbIkCxcupEePHgBERERQvnx5IiMjKVeuHD4+Pty6dYvt27er53399dds3ryZc+fOARmZZP7+/vj7+7N9+3aaN29OZGQkZcuWzTI2Pz8/DAwMmDdvnrpv//79NGjQgIcPH2JqavoqHokQQgghhMhBbmNFUpNMCCGEEHlKyNlYAjZGEBufBEDyP1EkJydjWLRijudUqlRJ/dzR0RGAuLg4ypUrR2RkJG3atNFr7+npycyZM0lPT1czxDKFh4dTtGjRbANkAKdOneL06dMsW7ZM3afT6dBqtURHR+Pq6vp8NyyEEEIIIV4LCZIJIYQQIs8IORtLn6UneDwNXjEyAWDkurM4FC2OdwXHLOcZGRn9r72iABnTJl+EmZnZU48nJCTQu3dv+vfvn+VY8eLFX+iaQgghhBDi1ZMgmRBCCCHyhHStjoCNETxZJ8IoX2EUQxOSrp4iYKMTTd0cMNAoue7X1dWVsLAwvX1hYWGULVs2SxYZZGSlXbt2jQsXLmSbTebh4UFERATOzs65HoMQQgghhHjzJEgmhBBCiDzhSPQddYrl4xRDY6xrfszd0CAuGhiyJtSGYuZpnDt3jsaNGz+z38GDB1O9enXGjx9Phw4dOHjwIHPmzOGnn37Ktn2DBg2oX78+H3/8MdOnT8fZ2Znz58+jKAre3t4MGzaMWrVq0a9fP/z8/LCwsCAiIoIdO3YwZ86c//wchBBCCCHEqyFBMiGEEELkCXEPsgbIMtl4foaiMeDevmV8vm0OhQs78sUXX+SqXw8PD3777TdGjx7N+PHjcXR0ZNy4cfj6+uZ4zurVqxkyZAgdO3bk4cOHODs7M3nyZAD69+/Phx9+yIULF6hXrx46nY7SpUvToUOH57pfIYQQQgjxesnqlkIIIYTIEw5evk3HBYee2W5Fz1rULm3/GkaUPS8vL9zd3Zk5c+YbG4MQQgghhPif3MaKNK9xTEIIIYQQL6xGSTscbUzJqdqYAjjamFKjpN3rHJYQQgghhHhHSJBMCCGEEHmCgUZhTGs3gCyBssztMa3dnqto/6t29+5dunTpQr58+TA3N6d58+ZcvHhRr83q1aspX748JiYmODk5ERgYqHf8p59+okyZMpiamlKoUCE++eQT9ZhWq2XSpEmULFkSMzMzKleuzB9//PFa7k0IIYQQ4l0jQTIhhBBC5BneFRyZ28kDBxtTvf0ONqbM7eSBdwXHNzSy7Pn6+nLs2DE2bNjAwYMH0el0tGjRgtTUVACOHz9O+/bt+eyzzzhz5gxjx45l1KhRBAcHA3Ds2DH69+/PuHHjiIqKIiQkhPr166v9T5o0icWLF/Pzzz9z7tw5Bg4cSKdOndi7d++buF0hhBBCiDxNapIJIYQQIs9J1+o4En2HuAdJFLTKmGL5tmSQZdYk+/LLLylbtixhYWHUqVMHgNu3b1OsWDEWLVrEp59+io+PD7du3WL79u3q+V9//TWbN2/m3LlzrFmzhm7dunHt2jWsrKz0rpOcnIydnR07d+6kdu3a6n4/Pz8SExNZvnz567lhIYQQQoi3XG5jRbK6pRBCCCHyHAON8kaL8z/uyYBd5ruPkZGRGBoaUrNmTbWtvb09Li4uREZGqm3atGmj15+npyczZ84kPT2dpk2bUqJECUqVKoW3tzfe3t589NFHmJubc+nSJRITE2natKne+SkpKVSpUuWV3rMQQgghxLtIgmRCCCGEEC8o5GwsARsjiI1PUvfdiblLvmIPX0r/VlZWnDhxgtDQULZv387o0aMZO3YsR48eJSEhAYDNmzdTpEgRvfNMTExeyvWFEEIIId4nEiQTQgghhHgBIWdj6bP0BE/WrUhJ07I7Mo7WBvakpaVx+PBhvemWUVFRuLllLEDg6upKWFiY3vlhYWGULVsWAwMDAAwNDWnSpAlNmjRhzJgx2Nrasnv3bpo2bYqJiQkxMTE0aNDgld+vEEIIIcS7ToJkQgghhBDPKV2rI2BjRJYA2eMWnk7iww/b0LNnT+bNm4eVlRXDhw+nSJEi6hTLwYMHU716dcaPH0+HDh04ePAgc+bM4aeffgJg06ZN/PXXX9SvX598+fKxZcsWtFotLi4uWFlZMWTIEAYOHIhWq6Vu3brEx8cTFhaGtbU1Xbt2fQ1PQgghhBDi3SFBMiGEEEKI53Qk+o7eFMvsxMYnMXZMIMtmBtCqVStSUlKoX78+W7ZswcjICAAPDw9+++03Ro8ezfjx43F0dGTcuHH4+voCYGtry5o1axg7dixJSUmUKVOGFStWUL58eQDGjx9PgQIFmDRpEn/99Re2trZ4eHjw7bffvtL7F0IIIYR4F8nqlkIIIYQQz2l9+HUGrAx/ZrtZn7nTxr3IM9sJIYQQQohXJ7exIs1rHJMQQgghxDuhoJXpS20nhBBCCCHePAmSCSGEEEI8pxol7XC0MUXJ4bgCONqYUqOk3esclhBCCCGE+A8kSCaEEEII8ZwMNApjWmesUPlkoCxze0xrNww0OYXRhBBCCCHE20aCZEIIIYQQL8C7giNzO3ngYKM/pdLBxpS5nTzwruD4hkYmhBBCCCFehKxuKYQQQgjxgrwrONLUzYEj0XeIe5BEQauMKZaSQSaEEEIIkfdIJpkQQgghxH9goFGoXdqeNu5FqF3aXgJkQohXzsvLC39//1fW/5UrV1AUhfDw8Fd2DSGEeBtJJpkQQgghhBBCCFWxYsWIjY0lf/78b3ooQgjxWkmQTAghhBBCCCGEysDAAAcHhzc9DCGEeO1kuqUQQgghhBBC5DFarZavv/4aOzs7HBwcGDt2rHps+vTpVKxYEQsLC4oVK0bfvn1JSEgA4P79+5iZmbF161a9/tauXYuVlRWJiYlZpluGhoaiKAq7du2iWrVqmJubU6dOHaKiovT6+O677yhYsCBWVlb4+fkxfPhw3N3dX+VjEEKIl0qCZEIIDA0N+eijj970MIQQQgghRC4tWrQICwsLDh8+zNSpUxk3bhw7duwAQKPRMHv2bM6dO8eiRYvYvXs3X3/9NQDW1ta0atWK5cuX6/W3bNky2rZti7m5eY7XHDFiBIGBgRw7dgxDQ0O6d++ud/6ECROYMmUKx48fp3jx4sydO/cV3LkQQrw6EiQTQrxyfn5+KIoUshZCCCGEeFHpWh0HL99mffh17j9KpWKlSowZM4YyZcrQpUsXqlWrxq5duwDw9/enYcOGODk50ahRI7777jt+++03tS8fHx/WrVtHYmIikJFdtnnzZnx8fJ46hgkTJtCgQQPc3NwYPnw4Bw4cICkpCYAffviBHj160K1bN8qWLcvo0aOpWLHiK3oaQgjxakiQTAghhBBCCCHeYiFnY6k7ZTcdFxxiwMpwImLvcynVjpCzsWobR0dH4uLiANi5cyeNGzemSJEiWFlZ0blzZ27fvq0GxVq0aIGRkREbNmwAYPXq1VhbW9OkSZOnjqNSpUp61wPUa0ZFRVGjRg299k9uCyHE206CZEK8IwYPHoypqSmKoqDRaLCzsyMuLg5bW1uqVKmi19bR0RFnZ2e9fffv36dEiRIoioKBgQHt27dXj2m1Wry8vDA0NFSPV65cWe/catWqYWBggKIoWFpaMnPmTABmzpzJL7/8AoCiKCiKgpeX16t5CEIIIXKtSZMm6u/lzP87AgMDAfjxxx9RFIUaNWqg0WjUNp06dXrDoxbi/RNyNpY+S08QG5+ktz8xDfosPaEGyhRFQavVcuXKFVq1akWlSpVYvXo1x48f58cffwQgJSUFAGNjYz755BN1yuXy5cvp0KEDhoZPX9fNyMhI/TxzloBWq305NyqEEG8BCZIJ8Q4IDw9n+vTpeHt7s2/fPn7//Xc++OCD5/qjZffu3bi6uhISEsJHH33E77//zuTJkwEYOnQoe/fuZfTo0Rw4cICFCxfqBclq1KjBhQsXmDVrFrt27cLLy4uBAweyY8cO/Pz81Hpnp06d4tSpU6xcufLlPgAhhBDPLT4+nurVqzN16lT69+8PwJAhQ9SpUwBHjx6lXr16fPfdd2g0GpYtW8a9e/fe0IiFeP+ka3UEbIxA95Q2ARsjSNf+r8Xx48fRarUEBgZSq1YtypYtyz///JPlPB8fH0JCQjh37hy7d+9+5lTLZ3FxceHo0aN6+57cFkKIt93T3yoQQrxUV65coWTJkpw8efKlrPSTkqZlycErbFobAoD/oCHUrVsXgI8//vi5+sqfPz8hIRn9NGvWjOLFizN9+nRq1arF9OnTMTIyYsiQIZibm1O7dm26desGwMGDB4mKiuL48eN4eHgA0KhRI+zs7BgzZgwHDhzAzs4O0E/RF0II8XolJKUxcNVJYu4+ong+M/bsO4il6f/+FGzevDnNmzdnzJgxFC9eHIDSpUuzd+9eABITE5k4cSKzZs1izJgxb+QehHjfHIm+kyWD7HE6IDY+iSPRd9R9zs7OpKam8sMPP9C6dWvCwsL4+eefs5xbv359HBwc8PHxoWTJktSsWfM/jfWrr76iZ8+eVKtWjTp16rBq1SpOnz5NqVKl/lO/QgjxOkkmmRCviK+vL23btn1l/U/aEkG5UVsZvzmS4xoXNKaWNGxQD5v8DnTp0oXo6Ojn6q9ixYp4eXnh7+8PQLVq1bh79y516tRh06ZNaLVarK2tKVeuHMOGDVMzDbZt2wZA1apV9abt3L17l2vXrr3UexbPJ3O5dsn6ECLvcHJyUqerv0wfztlHhbHb2BEZR9SNB+yIjKNosx5oNAbq7+3mzZsDcPjwYfW8x6fHN2vWDIAzZ8689PEJIbIX9yDnAFlO7SpXrsz06dOZMmUKFSpUYNmyZUyaNCnLOYqi0LFjR06dOvWfs8ggIzPtm2++YciQIXh4eBAdHY2vry+mpqb/uW8hhHhdJJNMiDxo0pYI5v35vyCYxtCYIl8tJyE8hMTIP1mxchVLly5VgyQ6nX6Sfnp6eq6vZWxsTMuWLYmLi2PGjBls2LCBadOm8dNPP3Hz5k3u3r0LwJYtWzA2NtY7t0CBAv/hLsXz8vLywt3d/ZW8wBZC5F0fztnH6Wv3s+yP/3MxAAXdatH746akpaUxadIkkpOT1TYWFhbq5xpNxnurz/N/iBDivylolX2AyeHzyVnarVu3Tt0eOHAgAwcO1GvTuXPnLP1MmTKFKVOmZNnv5OSk9/ejl5dXlr8n3d3ds+wbNWoUo0aNUrebNm2apQ6uEEK8zSST7D1Sv359tcCm0Ofl5cVXX32Fv78/+fLlo1ChQixYsICHDx/SrVs3rKyscHZ2ZuvWrUDGC4QePXpQsmRJzMzMcHFxYdasWWp/Y8eOZdGiRaxfv159hz40NFQ9/tdff9GwYUPMzc2pXLkyBw8e1BvP/v37qVevHmZmZhQrVoz+/fvz8OFDIGOK5eiODbl3YCX/bgokZvonXJvbjaTLR7Eo54nG1II0Leh0OkaNGoWVlRW3b9/m9u3bdOzYEUdHR27dukVMTAwrVqxQr3ngwAH27t3LrFmzUBSFtWvXYm1trQbaNBoN48eP59SpU0ydOpWEhARsbW3VovxnzpyhcePG3L17F39/f1q1akXDhg1p0qQJBgYGr+pLJ4QQ4ikSktKyDZAlRp8AwKysJ2atR/L1t6O5fPny6x6eEOIZapS0w9HGlJz+glcARxtTapS0e53DylZiYiLTp0/n3LlznD9/njFjxrBz5066du36pocmhBC5JkGy98iKFSvUuiK5ZWho+N5kAy1atIj8+fNz5MgRvvrqK/r06cOnn35KnTp1OHHiBB988AGdO3cmMTERrVZL0aJF+f3334mIiGD06NF8++23/Pbbb0BG4eP27dvj7e1NbGwssbGx1KlTR73WiBEjGDJkCOHh4ZQtW5aOHTuSlpYGwOXLl/H29ubjjz/m9OnTrFq1iv3799OvXz8Alhy8gg54cHQ9JkXdcPSdhaGNA7fWTuDGim8xdiiDhWt9ACIiIqhXrx7Xrl1jwoQJmJubqynvNjY2dO7cmSNHjgCQnJyMra0t3t7etG7dGkCdegnQt29f1q5dy8KFCxk6dCgAf/zxB4cPH8bOzo5vv/2Wvn370qFDB8qXL0/16tVp37497dq1w83NDYDvv/+eqKgo/v3331f0VXx/+fr66gU5FUXhypUrQEYB32rVqmFubk6dOnWIiopSz7t8+TJt2rShUKFCWFpaUr16dXbu3KnXt5OTExMnTqR79+5YWVlRvHhx5s+f/zpvT4g8Yf78+RQuXDjLoilt2rShe/fuufp5e9yVK1dQFIXw8HB1371797K88XL27FmaN2+OpaUlhQoVonPnzurv2YGrTmbbt3GxjN/Ljy4e5N6h1bjVaqj+HyaEeHsYaBTGtM74eX0yUJa5Paa1GwaaN/9GuKIobNmyhfr161O1alU2btzI6tWradKkyZsemhBC5J7uHRMfH68DdPHx8W96KO8EAwMDXf78+d/0MF6JtHSt7sClf3XrTl7TValRR+dZt+7/jqWl6SwsLHSdO3dW98XGxuoA3cGDB7Pt78svv9R9/PHH6nbXrl11bdq00WsTHR2tA3QLFy5U9507d04H6CIjI3U6nU7Xo0cPXa9evfTO27dvn06j0egePXqkG7XujM7AuqDOonxDXYlhm3Qlhm3S2bcYpCOjdmvGh8ZQV6JiDR2gu3z5sq58+fI6RVF0iqLomjVrpnNwcNCVLl1a17JlS93gwYN1BgYGuoYNG+pMTEx0gE6j0ejatWun0+l0uj179ugAnYWFhdq/RqPRTZ06VR3fw4cPdfXr19cZGBioxx0dHXV//PGH2sbNzU2nKIoO0DVo0CCXXyWRW/fu3dPVrl1b17NnT11sbKwuNjZWt3PnTh2gq1mzpi40NFR37tw5Xb169XR16tRRzwsPD9f9/PPPujNnzuguXLigGzlypM7U1FR39epVtU2JEiV0dnZ2uh9//FF38eJF3aRJk3QajUZ3/vz5N3GrQry17ty5ozM2Ntbt3LlT3Xf79m11X25+3kxMTHR169bVDRgwQGdtba0DdKNGjdIlJCTofH19dZaWljpAN3nyZJ1Op9PdvXtXly9fPl2pUqV05ubmOjs7O52jo6PO09NTp9PpdB/M2Ksr+GmAzqSIm04xsdBpTK10ZqWr6wr3WqAzLVlV7/+OGjVqqL/DixQpogN0/fv31+l0Ot2VK1d0derU0QE6AwMDnZubm27z5s2v8ekK8X7beuYfXa2JO9W//UoM26SrNXGnbuuZf9700IQQIk/IbaxIMsnyEENDQ+zt7bG3t1czRYoXL67WBgkPD8fKyko9ZmxsTHBwsHr+k9MtnZyc0Gg0NGrUSD3H3NycCxcuAGBtbU16ejr//vuvevz3338nPDwcCwsLvSLt9erVe63P4r8KORtL3Sm76bjgEANWhhMRe59LqXaEnI0FwMDAAHt7eypWrKieU6hQIQDi4uIA+PHHH6latSoFChTA0tKS+fPnExMTk6vrP77Ko6Ojo16/p06dIjg4GEtLS/WjWbNmaLVaoqOjKWFnDoBRASe1D4sKDQHI32Y4JYZtosTQdXwxYioACQkJnD17ltTUVAICArh+/TrJycncuHGDbdu2ERMTQ1paGrt376ZWrVoMGDCA9PR0Vq9erTfma9euodPpcHV1ZeTIkWo2GYC5uTl79+4lOTmZxo0bY2FhgaenJ3fu3FFrlp07dw6tVotOp9PLgBD/TbpWx8HLtwmNTiApXcHMzAwHBwccHBzUaa4TJkygQYMGuLm5MXz4cA4cOKAuvFC5cmV69+5NhQoVKFOmDOPHj6d06dJs2LBB7zotWrSgb9++ODs7M2zYMPLnz8+ePXte+/0K8TbLly8fzZs3Z/ny5eq+P/74g/z589OwYcNc/7wdPXqU/Pnzq/WFJk6cqGY2Z/7+nDhxIomJiXz//fckJCTQoUMHjh8/zs6dO3F2diYsLIwLFy5QPJ8Z2tQkrKu3xbHrDAp9NgEUhVtrJ1Dw0zGUGLaJIl9kTJu/f/8+mzZtIjIyktq1a1OiRAkCAwMB+PLLL7GwsOD06dNcuHCBKVOmYGlp+eofqhACAO8Kjuwf1ogVPWsx6zN3VvSsxf5hjfCu4PimhyaEEO8UCZLlMXfu3EGj0TB//nwaN27M33//Td26dQGoVasWCQkJDBgwQC3A2a1bN+Lj43PsT6fTcfDgQWbMmMGAAQN49OgRjRo1AjJqVGk0GmxsbNi1axe7du2iVatWNG7cmEePHjF27Fh+++03/Pz8KFy48Ku/+Zck5GwsfZaeyLKcdmIa9Fl6Qg2UKYqCkZGRejwzwKjValm5ciVDhgyhR48ebN++nfDwcLp160ZKSkquxpBTv5AR1Orduzfh4eHqx6lTp7h48SKlS5emc20nFEDRGGbpQ9FkBEU0CrSrWkyv3++//55Zs2YxbNgw9uzZQ3h4OM2aNcv1mDOZmZnleMzAwIAdO3awdetW3Nzc+OGHH3BxcXnulTZF7mQX7P3t2DX1ezjT04KyCQkJDBkyBFdXV2xtbbG0tCQyMjJLwPfxPhRFwcHBQe1DiPddZrB6ffh1qjf+kNWrV6vF75ctW8Znn32GRqPJ9c9b4cKFGTlyJCVLlgQyFlDJnz8/PXv2pHTp0kBGQOv06dOsX7+etLQ0Zs+eTbVq1ahXrx7Hjh0DYO/evczoUAULF0/MXepglK8wxoVKYd98AKm3rpD6r/51hwwZQsuWLSlbtiwBAQFcvXqVS5cuARATE4OnpycVK1akVKlStGrVivr167/S5yqE0GegUahd2p427kWoXdr+rZhiKYQQ7xpZ3fIt9yglnYlbIrhyOxHt/68ec+PGDQwMDOjZsyeFCxfm0KFDBAcHk5yczNChQ5k6NSODyMvLi5o1a+Lj48OmTZtyvMapU6coW7YsAKtWrSI2NuMFdoUKFdRAUWbgDDJeVFtZWTFmzBgAPv3001dy769CulZHwMYIdE9pE7AxgqZuDk/tJywsjDp16tC3b19135MFj42NjV9oBTAPDw8iIiKeuhKQpenTf3R71iuJsaF+DDwsLIw2bdrQqVMnICN4duHCBbVeWG7HXKlSJXbt2kVAQEC2xxVFwdPTE09PT0aPHk2JEiVYu3YtgwYNemq/4vlkBnuf/F5+mJxGn6UnmNvJg8z1sJ4WlB0yZAg7duxg2rRpODs7Y2ZmxieffJIlePp4H5n9PFl3SYj3UcjZWAI2RqhvvOjS8pGQlMp3c5fS6+MP2LdvHzNmzABy/nlLTk7m4OXbxD1IQqsDx/9/4ylzNUkbGxs1szk1NVW9dlxcnFp7TPfYCnOZP+f58+fH0tSQUsbxHP5jHin/RJH+6D78f9v0+7eggBPlHCy5Ts4B9XLlytG/f3/69OnD9u3badKkCR9//LFeeyGEEEKId4Fkkr3Fei4+iuvoEJYcimHfxX/R6UAxNuOLZSfUNpnF4DODYI8HLmrUqIGiKJw7dy7HayiKogbIIOMP6me98G3fvj33799Ho9Hg6OjI119//UL39yYcib6TJYPscTogNj6JI9F3ntpPmTJlOHbsGNu2bePChQuMGjWKo0eP6rVxcnLi9OnTaqH6x1/YPM2wYcM4cOAA/fr1Izw8nIsXL7J+/Xq1cD+AjZkRdUrb8eQbiIoCveuX5JsWbjypTJky7NixgwMHDhAZGUnv3r25efNmljEfPnyYK1eu8O+//2b7vfDNN99w9OhR+vbty+nTpzl//jxz587l33//5fDhw0ycOJFjx44RExPDmjVruHXrFq6urrm6d5E7OQV7FQMj0GV8zQI2RpCufVo4OENYWBi+vr589NFHVKxYEQcHB7XgvxDi6bLLTFYMjTErU4fpc39l7Ix5uLi44OHhAWT/83bpr2h+O3ZNzQhNTddy5vp9Qs7GqgvnaLVaNVD9eBF/rVaLhYUFFhYWHD9+nFOnTqkfFy9e5IMPPgDgn9/GYaF7hJ33Vzh2no5D54wplLr0NCoVtWZh1xrA0wPqfn5+/PXXX3Tu3JkzZ85QrVo1fvjhh1fxWIUQQggh3hgJkr2lei4+yo6I7Kcy7YiIo+fio9keex2WLFnC/v37qVevHg8ePOD777/HweHpmVdvi7gHOQfInqdd7969adeuHR06dKBmzZrcvn1bL6sMoGfPnri4uFCtWjUKFChAWFhYrq5dqVIl9u7dy4ULF6hXrx5VqlRh9OjRWaa0NnFz4Pz45oxq6UqX2iUAmPO5R7YBMoCRI0fi4eFBs2bN8PLywsHBgbZt2+q1GTJkCAYGBri5uVGgQIFsa6yVLVuW7du3c+rUKWrUqEHt2rVZv349hoaGWFtb8+eff9KiRQvKli3LyJEjCQwMpHnz5rm69/eZTqejV69e2NnZoSgKtra2equLPi6nYK+hTUGSY6NIjb/JtdibRPxzTz0WHByMra2tuj1z5kzc3d0pU6YMa9asUaf1fv755+qLYl9fX73vES8vL70xzZw5k5kzZ/6HuxYi73paZrJ5eS8SLx9l+ZLFdPz8c3X/kz9vTVq3IzE5lYfJaXrnp6Tp6LP0BHsv36NWrVrEx8dz8+ZN9u7dy8iRI/XatmzZkkePHjFixAju3r2LoihcvnyZCRMmYGpqyu3bt4mKimL1/EAuBw+jRb2qFLXI+Bmf9VkVNvTLfU3RYsWK8cUXX7BmzRoGDx7MggULcv/AhBBCCCHyAJlu+RZ6lJKeY4BMl/IIyAiUPUpJ5+DBgwC0atWK1atXM2bMGHW65ZEjR9DpdFSoUOGFx6Ioit4Ujkyenp7s3bsXyMhYezKL6m1V0Mo02/0On0/O0i67bJrHn0VQUBBBQUF6xydNmqR+XqBAAbZv3/7UPgBsbW2z7KtevXq252Z6fGw96pUCYNwTfTg5Oen1a2dnpxaBzknZsmXV76mc+gFo0KBBtkE/W1tbQkJCnnoNkb2QkBCCg4MJDQ2lVKlSfPLJJzm2zSmIa12jHf9uns4/C/uiS0vmYsB09ViHDh1o0aIF//zzj94506dPp3v37tSpU4f8+fMzbNgw7t+/D8CsWbPQ6XS4u7tne72ePXvSq1cvdVtRFNauXZsl+CrEu+hpmcmmJSphYGZF0r9/41a3hbr/yZ83A/e2GBeKzbYPyMgIXbDwF6pV9SAwMJCQkBCmTp2qZogBDB8+nGXLlnHo0CEaNWpEamoqBQoUwNraGp1OR758+bC3t2f+/PmMcXSkY7EHnPpxMQBmxga5vl9/f3+aN29O2bJluXv3Lnv27JEsYSGEEEK8cyRI9haauCXiqcf//sEH23qdKV9zAv/88w/Vq1fH19eXL774gmnTppGWloaDg4P6bvPSpUtfeCwWFhbcvXuX33//nWLFiuHh4YGzszM1atTgww8/JD4+nlOnTql1U952NUra4Whjyo34pGzf/VcABxtTapS0e91DE++5y5cv4+joqE6hNjTM+ddzTsFeI7siOP7/NCqAjj61mD16oLptZmZGwYIF0el0jB07FsgIgu7evVuvny+//FJvOzMo+3iQ9fEpX0K8j56WcawoGop+mRGIMsr3v0zrx3/eDl6+TccFh3Bwa6Z3rklhF4wLllKn/z8wLYSDgwP+/v5qJqdOp1OnQ2bWJs1clCU9PR1LS0s++OADDAwMUBSFlStX0r9/fypUqICLiwuzZ8/Gy8vrue43PT2dL7/8kmvXrmFtbY23t7daa00IIYQQ4l0hQbK30JXbiTke05haoktP4862OdwBihQpomb+HDp0iHr16ql/tBoZGREUFISNjc0LjyUwMJBevXrRvn17AH777TeMjIxYvXo1q1evBsDExERvufu3mYFGYUxrN/osPYECeoGyzPJeY1q7yWpB4rXy9fVl0aJFQEY2VokSJXByctJrk5yczIgRI1ixYgX37t1DsSuOZb0umBavhC4thdhgf0yKumLv/RUKYJt2hw/cnZg1axbdu3cnODgYf39/7t27p9fvvHnz+O6777h9+zatWrViwYIF6u8MX19f7t27l2MGopOTk/rCPXO8H330EQAlSpRQs+KOHDlCtWrV1PNmzpzJjBkziI6OzjMBdiGelFOwOrftcgqyPZnZHPcg6ZmZzZnTOHPSpEkTIiL034B7/PzsMoafzHKW+mNCCCGEeB/Iq5O3kJO9ec4HFQ3F/VdRYtgmRq49zbVr1zAwyJgu4e7uzoMHD9DpdOh0OlJSUvD19VVP/fPPP/X+4L1y5UqWwuxnzpzRa9OjRw/S09PVPj/99FMuX76sbut0OpKSkujQocPLufnXwLuCI3M7eeBgo//CxcHGlLmdPPCu4PiGRibeV7NmzWLcuHEULVqU2NjYbKcv9+vXj4MHD7Jy5UpOnz7NZx0+5eZvY0i7cx3F0Jj8rYeQcHYXjy4eQqdNJ2nHLJo2bUr37t1zvO6lS5f47bff2LhxIyEhIZw8eTJLbb3cyhxzUFCQeg9OTk40adIky7TkoKAgfH19JUAm8rTMzOSc3lJRAMenZCb/1yCbEEIIIYR4+SST7C30bQs3lhzKWjA9u3bixXhXcKSpmwNHou8Q9yCJglYZL2Qkg0y8Tulanfo9eDtZg4GBQbaLYMTExBAUFERMTIy6gMPCaeM4vn8PcRdCoZYPxoVKYVuvM3dCfqB5m3Ycu/UPC3Zve+r1k5KSWLx4MUWKFAEyMkVatmxJYGDgcy/GkbkKn62trd65fn5+fPHFF0yfPh0TExNOnDjBmTNnWL9+/XP1L8Tb5r9mJsv0fyGEEEKIt4+8jf8WMjM2oKlbwae2aepW8LkK7oqsDDQKtUvb08a9CLVL20uATABZV3B8XleuXEFRFLVmV2hoKIqiZJnmGHI2lrpTdtNxwSEGrAxn8cGrxMYnEXI2axHvM2fOkJ6eTtmyZbG0tFQ/zh4/hGfBdFb0rMWsz9zZNH8K1d3Ls3FFEL/++iv29vZPHWvx4sXVABlA7dq10Wq1REVFvfD9P6lt27YYGBiwdu1aIGOVzYYNG2aZTipEXvRfMpMzg2xAlmw0mf4vhBBCCPFmSCbZW2pBl+r0XHxUb5XLEl9vADICZAu6VH9TQxPinbZmzRqMjIxeWn916tQhNjZWrzZgyNlY+iw9kSV7JF2ro8/SE8zt5KG3PyEhAQMDA44fP65Or85kaWmJg0NGMOzGjRtcvHABAwMDLl68iLe3N5BRW+z06dMv7Z6eh7GxMV26dCEoKIh27dqxfPlyZs2a9UbGIsSr8F8ykzODbAEbI/RWynSwMWVMazeZ/i+EEEII8ZpJkOwttqBLdR6lpDNxSwRXbifiZG/Oty3cJINMiFfIzu7lTm0yNjbWm36YrtURsDEi2+lVmQI2RmD82HaVKlVIT08nLi6OevXqqftTUlIwNv5fy+7du1OxYkV69OhBz549adKkCa6urjleJyYmhn/++Uedwnno0CE0Gg0uLi7PfZ+QsVhIenp6lv1+fn5UqFCBn376ibS0NNq1a/dC/QvxtsrMTH4RMv1fCCGEEOLtIUGyt5yZsQHj21Z808MQ4r3h5eWFu7s7M2fOxMnJiV69enHp0iV+//138uXLx8iRI+nVq5fa/siRI/Tu3ZvIyEgqVKjAiBEj9PoLDQ2lYcOG3L17F1tbW9bvP8WpX74h+VoEOm0ahtaFyNewm9o++dZVTv0RTHL0CQ4eOMDx48cJDg7Gx8eH5s2b4+bmRv369Vm0aBHp6eksWbKESpUq0aZNG06ePImNjQ1WVlY0adIEHx8fWrZsqa6cCRmrZ+7ZswfICOBVqVKFxMREFEVBq9XSokWL565HlsnJyYldu3bh6emJiYkJ+fLlA8DV1ZVatWoxbNgwunfvjpmZ2Qv1L8S76r8E2YQQQgghxMsjNcmEEOIpAgMDqVatmrryY58+fdSaXQkJCbRq1Qo3NzeOHz/O2LFjGTJkyFP7mzRyCLr0NAr5TMGx+4/YevmiGGcEjXTadG4uH45iYIRTWVc6duxI9+7dSUtLIygoCGdnZ44dO8b06dMB8PDwwNHRkYYNG3L69GkmT57MwYMHsbS05Pz589y6dYuEhATat29PxYoVsbKyIjY2ljp16pCeno5Wq6V06dKYmZmRnJyMnZ0dUVFRpKSkvPCz2rFjB8WKFaNKlSp6x3r06EFKSspTV9sUQgghhBBCiDdJMsmEEO+9x1eZvP8oFZ3uf5MhW7RoQd++fQEYNmwYM2bMYM+ePbi4uLB8+XK0Wi2//PILpqamlC9fnmvXrtGnT58cr3U3LhaTIu4YF3ACwMg2I2vLtFgF0hPvkRj5J/k//JqlX9TNklni7u7OjRs3iImJUadZLl26FENDQ1JTU1GUjOlZQUFB2Nrasm7dOj744AN8fX0pVaqUXl0yFxcXnJycCAsLU89LSUnB1taW0NBQPvjgA4KDg/WuHxoaqrd95coVve3WrVvTunXrbO/7+vXrVKxYkerVpZ6iEEIIIYQQ4u0kQTIhxHst5GysXtHsG7H3iT12jeb/v8pkpUqV1LaKouDg4EBcXMaCGpGRkVSqVAlT0/+tbFe7du2nXu/rQf580acPSVdOYlrCHXOXOhgXLAlAStxfmBQrT2E7S2qUzL42WsWKFfXqkJ06dYpLly5hZWWl1y4pKYnLly/nOI4XPe95JSQkcOXKFebMmcN333330voVQgghhBBCiJdNgmRCiPdWTqtMPkxOo8/SEzxKTc+y0mVm7a4X1atXT8xKVcH/+yCSrpwk9tDv5GvUA+uqrdEYmgAwprVbjkW7LSws9LYTEhKoWrUqy5Yty9K2QIECOY7jRc97Xv369WPFihW0bdtWploKIYQQQggh3mpSk0wI8V7KzSqT9xJT0epybuHq6srp06dJSkpS9x06dOiZ1+7cpBrLAkdSyXc81jXaknBqGwB2xZyxunuJxi75c30fHh4eXLx4kYIFC+Ls7Kz3YWNjA2QU6H9y1cncnPcyBAcHk5yczKpVqzAwMCA4OBhbW9uX1j9kTPtUFIXw8PCX1qeiKKxbt+6l9SeEEEIIIYR4+0mQTAjxXjoSfUedYpkdHRmBtJjbiTm2+fzzz1EUhZ49exIREcGWLVuYNm3aU6/r7+/Ptm3bcLFIYnZTW4o9iqZGlYqs6FmL4ysCSU16yGeffcaxY8e4ePEiS5YsURcKyI6Pjw/58+enTZs27Nu3j+joaEJDQ+nfvz/Xrl0DMladPH36NFFRUfz777+kpqayePFidDrdU8/LK4oVK0ZsbCwVKlR400MRQgghhBBC5GESJBNCvJfiHuQcIHvcg+S0HI9ZWlqyceNGzpw5Q5UqVRgxYgRTpkx5an/p6el8+eWXuLq60rJFczwqubF+eRC1S9tTsEB+du/eTUJCAg0aNKBq1aosWLAgy5TPx5mbm/Pnn39SvHhx2rVrh6urKz169CApKQlra2sAevbsiYuLC9WqVaNAgQKEhYVhYGDAJ5988tTz8goDAwMcHBwwNJQKAkIIIYQQQogXp+h0T5lLlAfdv38fGxsb4uPj89wLPSHE63Pw8m06Lnj21MgVPWtlWWUyr/P19WXRokV6+6Kjo7l69SpDhw7l1KlT2NnZ0bVrV7777js1+OTl5aVmay1ZsgQjIyP69OnDuHHj1BUy7969y4ABA9i4cSPJyck0aNCA2bNnU6ZMGSBj+qW/vz/37t1Trz137lymTZvG33//TcmSJRk5ciSdO3dWj58/fx4/Pz+OHTtGqVKlmD17Nk2bNmXt2rW0bduWK1euULJkSU6ePIm7uzsA586dY9iwYfz555/odDrc3d0JDg6mdOnSHD16lG+//ZaTJ0+SmpqKu7s7M2bMwMPDQ72moihq/0IIIYQQQoi8LbexIskkE0K8l2qUtMPRxpTsy+ODAjjamOa4ymReNmvWLGrXrk3Pnj2JjY0lNjYWIyMjWrRoQfXq1Tl16hRz587ll19+ybIi5aJFizA0NOTIkSPMmjWL6dOns3DhQvW4r68vx44dY8OGDRw8eBCdTkeLFi1ITU3Ndixr165lwIABDB48mLNnz9K7d2+6devGnj17gIzMu7Zt22Jubs7hw4eZP38+I0aMeOr9Xb9+nfr162NiYsLu3bs5fvw43bt3Jy0tIyvwwYMHdO3alf3793Po0CHKlClDixYtePDgwX95rEIIIYQQQog8TuamCCHeSwYahTGt3eiz9AQK6BXwzwycPW2VybwmXavjSPQd4h4kUdDKFCNjY8zNzXFwcABgxIgRFCtWjDlz5qAoCuXKleOff/5h2LBhjB49Go0m4z2VYsWKMWPGDBRFwcXFhTNnzjBjxgx69uzJxYsX2bBhA2FhYdSpUweAZcuWUaxYMdatW8enn36aZVzTpk3D19eXvn37AjBo0CAOHTrEtGnTaNiwITt27ODy5cuEhoaqY50wYQJNmzbN8V5//PFHbGxsWLlypTpVtWzZsurxRo0a6bWfP38+tra27N27l1atWr3oIxavkE6no3fv3vzxxx/cvXsXGxsbfH19mTlzZq7Ozy7bUAghhBBCiCdJJpkQ4r3lXcGRuZ08cLAx1dvvYGPK3E4eeFdwfEMje7lCzsZSd8puOi44xICV4XRccIiTMXe5cvuh2iYyMpLatWur0yYBPD09SUhI0CvkX6tWLb02tWvX5uLFi6SnpxMZGYmhoSE1a9ZUj9vb2+Pi4kJkZGS2Y4uMjMTT01Nvn6enp9o+KiqKYsWKqQEygBo1ajz1fsPDw6lXr16Otdxu3rxJz549KVOmDDY2NlhbW5OQkEBMTMxT+xVvTkhICMHBwWzatInY2FguXLjA+PHjc33+k4s7hIaGoiiK3rRfIYQQQgghJJNMCPFe867gSFM3B70sqxol7d6ZDLKQs7H0WXqCJ4tPpqRp2R0ZR8jZ2HcmGJjJzMzsqce7du3K7du3mTVrFiVKlMDExITatWuTkpLymkYontfly5dxdHRUMxSfV+biDkIIIYQQQjyNZJIJId57BhqF2qXtaeNehNql7d+ZAFm6VkfAxogsATIAxcAIdFoCNkaQrtXh6uqq1hDLFBYWhpWVFUWLFlX3HT58WK+fzJpeBgYGuLq6kpaWptfm9u3bREVF4ebmlu0YXV1dCQsL09sXFhamtndxceHvv//m5s2b6vGjR48+9b4rVarEvn37cqyDFhYWRv/+/WnRogXly5fHxMSEf//996l9ijfH19eXr776ipiYGBRFwcnJCS8vL/z9/dU2Tk5OTJw4ke7du2NlZUXx4sWZP3++evzKlSsoikJ4eDhXrlyhYcOGAOTLlw9FUfD19QVAq9UyadIkSpYsiZmZGZUrV+aPP/54nbcrhBBCCCHeIAmSCSHEO+pI9B1i45OyPWZoU5Dk2Cj+jrnK9uMX6du3L3///TdfffUV58+fZ/369YwZM4ZBgwap9cgAYmJiGDRoEFFRUaxYsYIffviBAQMGAFCmTBnatGlDz5492b9/P6dOnaJTp04UKVKENm3aZDuOoUOHEhwczNy5c7l48SLTp09nzZo1DBkyBICmTZtSunRpunbtyunTpwkLC2PkyJEAetM+H9evXz/u37/PZ599xrFjx7h48SJLliwhKipKHeeSJUuIjIzk8OHD+Pj4PDP7TLw5s2bNYty4cRQtWpTY2Ngcg6SBgYFUq1aNkydP0rdvX/r06aN+zR9XrFgxVq9eDWRM542NjWXWrFkATJo0icWLF/Pzzz9z7tw5Bg4cSKdOndi7d++ru0EhhBBCCPHWkCCZEEK8o+IeZB8gA7Cu0Q4UDf8s7EuLGi6kpqayZcsWjhw5QuXKlfniiy/o0aOHGpDK1KVLFx49ekSNGjX48ssvGTBgAL169VKPBwUFUbVqVVq1akXt2rXR6XRs2bIlx/pgbdu2ZdasWUybNo3y5cszb948goKC8PLyAjKmya1bt46EhASqV6+On5+furqlqalptn3a29uze/duEhISaNCgAVWrVmXBggXqGH755Rfu3r2Lh4cHnTt3pn///hQsWDDXz1W8XjY2NlhZWalTJgsUKJBtuxYtWtC3b1+cnZ0ZNmwY+fPnV1dJfZyBgQF2dhmr1hYsWBAHBwdsbGxITk5m4sSJ/PrrrzRr1oxSpUrh6+tLp06dmDdv3iu9RyGEEEII8XaQmmRCCPGOKmiVfRAJwMiuCI6dAwFY0bMWTk72ODk5ceTIkaf2aWRkxMyZM5k7d262x/Ply8fixYtzPN/X11ed2papT58+9OnTJ8dzypUrx/79+9XtzOmZzs7OQMZUu8eniULGlMtt27Zl21+VKlWyZCN98sknettP9iderydXY9Xm4utRqVIl9XNFUXBwcCAuLi7X17x06RKJiYlZVk5NSUmhSpUquR+8EEIIIYTIsyRIJoQQ76gaJe1wtDHlRnxS9nXJyFjJs0ZJu9c9tOeydu1aLC0tKVOmDJcuXWLAgAF4enpSunTpNz008QqEnI0lYGOE/lThsxd5lJr+1POezFZUFAWtVpvr6yYkJACwefNmihQponfMxMQk1/0IIYQQQoi8S4JkQgjxjjLQKIxp7UafpSdQQC9QllnNa0xrt7d+oYIHDx4wbNgwYmJiyJ8/P02aNCEwMPBND0u8Ajmtxnr/URr3E1Je2mqsxsbGAKSn/y/w5ubmhomJCTExMTRo0OA/X0MIIYQQQuQ9EiQTQoh3mHcFR+Z28siSmeNgY8qY1m7PFXAIDQ19BSN8ti5dutClS5c3cm3x+jxtNdZMARsjaOrm8J+vVaJECRRFYdOmTbRo0QIzMzOsrKwYMmQIAwcORKvVUrduXeLj4wkLC8Pa2pquXbv+5+sKIYQQQoi3mwTJhBDiHeddwZGmbg56NZ5qlLR76zPI3napqak5Lkggnt/TVmPNFBufxJHoO//5WkWKFCEgIIDhw4fTrVs3unTpQnBwMOPHj6dAgQJMmjSJv/76C1tbWzw8PPj222//8zWFEEIIIcTbT1a3FEKI94CBRqF2aXvauBehdmn7dy5AptVqmTRpEiVLlsTMzIzKlSvzxx9/ABkZcIqisGvXLqpVq4a5uTl16tQhKipKr4+5c+dSunRpjI2NcXFxYcmSJXrHFUVh7ty5fPjhh1hYWDBhwgTGjh2Lu7s7v/76K8WLF8fS0pK+ffuSnp7O1KlTcXBwoGDBgkyYMEGvr5iYGNq0aYOlpSXW1ta0b9+emzdvqsd9fX1p27at3jn+/v7qqp8Af/zxBxUrVsTMzAx7e3uaNGnCw4cPX8LTfDOeuhpr9TYU7fOr2i40NJSZM2eqx69cuYK/v7/eOeHh4YwdOxb43+IO7u7u6vFRo0YRGxuLVqslODgYyPgaDxgwgPPnz5OSkkJcXBwhISHUr1//ZdyiEEIIIYR4y0mQTAghRJ43adIkFi9ezM8//8y5c+cYOHAgnTp1Yu/evWqbESNGEBgYyLFjxzA0NKR79+7qsbVr1zJgwAAGDx7M2bNn6d27N926dWPPnj161xk7diwfffQRZ86cUc+/fPkyW7duJSQkhBUrVvDLL7/QsmVLrl27xt69e5kyZQojR47k8OHDQEZAr02bNty5c4e9e/eyY8cO/vrrLzp06JDr+42NjaVjx450796dyMhIQkNDadeuXZ5elfNpq7G+SDshhBBCCCGel0y3FEIIkeeka3Xq9FFbY4WJEyeyc+dOateuDUCpUqXYv38/8+bNo1evXgBMmDBBLcg+fPhwWrZsSVJSEqampkybNg1fX1/69u0LwKBBgzh06BDTpk2jYcOG6nU///xzunXrpjcWrVbLr7/+ipWVFW5ubjRs2JCoqCi2bNmCRqPBxcWFKVOmsGfPHmrWrMmuXbs4c+YM0dHRFCtWDIDFixdTvnx5jh49SvXq1Z95/7GxsaSlpdGuXTtKlCgBQMWKFf/jU32z3pXVWIUQQgghRN4lQTIhhBB5SsjZWL2FCFJuXSUxMZFGjZvoTSNNSUmhSpUq6nalSpXUzx0dMxYsiIuLo3jx4kRGRqrBtEyenp7MmjVLb1+1atWyjMfJyQkrKyt1u1ChQhgYGKDRaPT2xcXFARAZGUmxYsXUABlkrKxoa2tLZGRkroJklStXpnHjxlSsWJFmzZrxwQcf8Mknn5AvX75nnvu2eldWYxVCCCGEEHmXTLcUQgiRZ4ScjaXP0hN6Bd51qRmf27YdxezfthMeHk54eDgRERFqXTJAr8i+omQEWrRa7XNd38LCIsu+J4v3K4qS7b7nuZZGo8kydTI1NVX93MDAgB07drB161bc3Nz44YcfcHFxITo6OtfXeBtlrsbqYKM/pdLBxpS5nTyeazVWIcSrcePGjTc9BCGEEOKVkSCZEEKIPCFdqyNgY0SWqXhG9sXAwIi0+7dYcOoRJUuVxtnZGWdnZ71sradxdXUlLCxMb19YWBhubm4vafT61/r777/5+++/1X0RERHcu3dPvV6BAgWIjY3VOy88PFxvW1EUPD09CQgI4OTJkxgbG7N27dqXPt7XzbuCI/uHNWJFz1rM+sydFT1rsX9Yo/c6QBYcHIytre2bHoZ4SyUlJWFvb4+iKOpH1apVSUpKQlEUdUp2pq+++gpFUfj5558B2L9/P8bGxuq5Go1GbyEMW1tbFEWhdOnSKIqCo6Mj5cqVQ1EUrl27pte3gYEBhoYyUUUIIUTeJf+LCSGEyBOORN/RyyDLpDExx7pGO+7sXshFnY41NW0obashLCwMa2vrLC8QszN06FDat29PlSpVaNKkCRs3bmTNmjXs3Lnzpd9HkyZNqFixIj4+PsycOZO0tDT69u1LgwYN1OmcjRo14vvvv2fx4sXUrl2bpUuXcvbsWXX66OHDh9m1axcffPABBQsW5PDhw9y6dQtXV9eXPt43IXM1VvH+URQFU1NTHj169KaHkmcUKVKEO3fu0KRJE1q2bMncuXM5ceIEn3/+Oebm5sTExOi1//XXjJViv/jiCwB19daePXtSqlQpxo8fz6xZs/jwww9p1KiRel50dDT9+vXD2NiYPn36UKZMGdq0acPx48cBWL9+PVqtlg8//PB13LYQQgjxSkiQTAghRJ4Q9yBrgCyTbb1OGJhbE3/odz5vNod8+Wzx8PDg22+/zdU0x7Zt2zJr1iymTZvGgAEDKFmyJEFBQXh5eb3EO8igKArr16/nq6++on79+mg0Gry9vfnhhx/UNs2aNWPUqFF8/fXXJCUl0b17d7p06cKZM2cAsLa25s8//2TmzJncv3+fEiVKEBgYSPPmzV/6eIUQb5eEpDQGrjpJzN1H2GnjuXPnDj4+PixduhQAf39/jI2N2bBhA19++SWzZ89m/vz59OrVi6SkJBITEylevDgAHTp0QKfT8ffff1O0aFEgY2ETRVHo0aOH3hTuDRs20KpVK3XbxMSEkydPqtu9e/cGYPXq1a/8GQghhBCviqLLy+vFZ+P+/fvY2NgQHx+PtbX1mx6OEEKIl+Tg5dt0XHDome1W9KwlWUjirbdp0yY6derE7du3MTAwIDw8nCpVqjBs2DAmT54MgJ+fH0lJSTRp0gR/f39WrVqFv78/f//9N3Xr1iUoKEhdhEKr1fLdd98xf/58Natw8uTJeHt7AxkLWQwaNIjVq1dz9+5dChUqxBdffME333wDwL179xgyZAjr168nOTmZatWqMWPGDCpXrvzanomRkRFpaWl6++bMmcOuXbv0phJbWFhw7dq193IK6odz9nH62n11+05oMA8O/5Fje51Oh6Io2Nvb8++//9KmTRs2bNjA9u3badq0KcWKFcsyZTKThYUFCQkJ2NraEh8fn6VO4ldffcWcOXNYtGgRXbp0QVEUbG1tuXv37su5WSGEEOIlym2sSDLJhBBC5Ak1StrhaGPKjfikLHXJIGMFRAcbU2qUtHvdQxPiudWrV48HDx5w8uRJqlWrxt69e8mfPz+hoaFqm7179zJs2DAAEhMTmTZtGkuWLEGj0dCpUyeGDBnCsmXLAJg1axaBgYHMmzePKlWq8Ouvv/Lhhx9y7tw5ypQpw+zZs9mwYQO//fYbxYsXz1IX79NPP8XMzIytW7diY2PDvHnzaNy4MRcuXMDO7vX8TP3555/UqVMHIyMjli9frt732rVrMTEx4ZtvviEkJIRDhw7h5OTEvXv3Xsu43hZPBsgA0hMyAlJmZWrj5FaJbp6l1WOZXzc7Oztu374NwNatW1EUhaZNmwIZ9cwApk6dmuV6FStWfOp4fvjhB+bMmcPAgQM5ePAgANOmTXuRWxNCCCHeGpJJJoQQIs/IXN0S0AuUKf//r6yAKPKSqlWr0rFjR4YMGcJHH31E9erVCQgI4Pbt28THx1O0aFEuXLhAWFgY3bp149KlS5QunREE+emnnxg3bpy60mCRIkX48ssv+fbbb9X+a9SoQfXq1fnxxx/p378/586dY+fOnerqrpn2799Py5YtiYuLw8TERN3v7OzM119/Ta9evV7pc3h8+uD2gQ30apJZWVmRkJBAamqqWhA+M8j36NEjTE1Nn9b1OyMhKY0KY7dl2Z8Ue4mbi/0xtCtKkZ4/c3ZsMyxN9d8D//HHH+nXrx9ffPEFP//8M5UrV1YXAvH29mbbtm3MmTOHL7/8Mttr55RJBlCiRAliYmIwMTEhJSXluVcMFkIIIV6X3MaKZHVLIYQQeYZ3BUfmdvLAwUb/hbGDjakEyESekK7VcfDybdaHX6dM5Rrs2ROKTqdj3759tGvXDldXV/bv38/evXspXLgwZcqUAcDc3FwNkAE4OjoSFxcHZPzR988//+Dp6al3LU9PTyIjIwHw9fUlPDwcFxcX+vfvz/bt29V2p06dIiEhAXt7eywtLdWP6OhoLl++/Eqfx4dz9lFh7DZ2RMYRdeMBAEmpWj6csw/IyCQzNDTUWzGxXbt2ACxevPiVju1tMnDVyWz3mzo6g7EZaXeucW1+bxp3G0Lfvn0pUqQILi4uAGrwK3M1yz/++N/0zE2bNgHQr18/mjVrxo8//oiPjw9WVla5Co6uXLkSgOTkZCpVqvTiNyiEEEK8JWS6pRBCiDzFu4IjTd0cOBJ9h7gHSRS0yphiaaBRnn2yEG9QyNlYAjZGqKu0JsYX4M7uJfy0eidGRkaUK1cOLy8vQkNDuXv3Lg0aNFDP1el02NraqlMMFUXJNrNHURTWrl1L27Zt9fZ7eHgQHR3N1q1b2blzJ+3bt6dJkyb88ccfJCQk4OjoqDfVM9OrrPuV3fTBTKev3VcDZQJi7ua82meRr1YQ+3N30u9e58jKWRwh4/sgM5gIGdl3MTExGBoa4uzsrO43NDTk+PHjeHp6sn37djV4amBgQM2aNZ85rtq1a2NoaEhaWpoacBNCCCHyMgmSCSGEyHMMNIoU5xd5SuZU4cfDWibFypOe/IhhAZOp6l4DAC8vLyZPnszdu3cZPHiw2tbIyIgLFy6o25kZPJCx2mnhwoUJCwvTu2ZYWBg1atTQa9ehQwc6dOjAJ598gre3N3fu3MHDw4MbN25gaGiIk5NTru4np2BcbiUkpeUYIMucTH362n3Mzc1JSEggLS1NzSZbs2YNAF26dHmha+dFxfOZqZl2TzI0NKRYv4ysuqauBVnQtXqWNlevXs2xbw8PD3V6a3aeVfstPT0dU1NTdXVMIYQQIi+T6ZZCCCGEEK9QulZHwMaILAtOGJhaYlTAiYfnQvnbpCTpWh3169fnxIkTXLhwQS+TTFEUChYsmOM1hg4dypQpUwC4fv06w4cPJzw8nAEDBgAwffp0VqxYwfnz57lw4QK///47Dg4O2Nra0qRJE2rXrk3btm3Zvn07V65c4cCBA4wYMYJjx4699OcBOU8fRGMI6ancP76JxMvH8eiYESi0tLQkICCAOnXq8Pfff2NjY/Pe1CMDmNGhyktt9zL8+eef1KxZE51Op1cLTwghhMjLJEgmhBBCCPEKHYm+o06xfFzipSOk3o4BnZbkAuVYunkv9vb22Nra4uDggIuLC35+fsybN4+UlBR16mNwcDCrVq0CMoJniqJgbW3NoEGDABgwYABz585Fo9HQokULNmzYgJWVFVOnTqVatWpUqVKF33//ndu3b1OkSBG++eYbNmzYQP369enWrRslS5akZcuWXL16lUKFCgHg7u7O2LFjAdRss48++ghFUXKdffa4nKYP2nh2BODuzp+59ccY4u7c46OPPiI5OZmxY8dy8OBBLCwsuHLlynNfMy+zNDWkUtGnL0hVqah1lqL9r1KDBg04cuQIBQoUYNSoUa/tukIIIcSrJEEyIYQQQrwxXl5e+Pv7q9tOTk7MnDnzjY3nVYh7kDVABmBarDxotTh0mYGRfTH27dtL/vz5KVWqFLGxsQDs3buXHj168NNPP6nndejQgcGDB1O+fHliY2OJjY2lY8eOjBkzBsgo6v/zzz9z9uxZWrRogY+PDx9//DEnT54kKioKgE6dOnH69Gnmzp3LL7/8wqxZs5g9ezbXr1+nRIkSjBkzhqVLl1KsWLEs4z569CgAQUFBxMbGqtvPo3g+s2z329bpQIlhm9SPuq19WLNmDTqdTv1ISEh4pbXS3lYb+tXLMVBWqag1G/rVe63jyfx6ZC4gIYQQQrwLJEgmhBBCCPEKFbTKflqgxsQC44IlSYo5A8DF8CMMHDiQkydPkpCQwPXr17l06ZLetEsAMzMzLC0tMTQ0xMHBAQcHB8zM/hd08vX1pWPHjjg7OzNx4kQSEhI4cuQIAD/99BPFihVjzpw5lCtXjrZt2xIQEEBgYCBarTZX91OgQAEANeMtc/t5vI3TB/OCDf3qcXZsM5q6FsTFwYqmrgU5O7bZaw+QCSGEEO8qCZIJIYQQ4qVLTU1900N4a9QoaYejjSnZrb9qUqwCyX+fwcHahHMnDtOuXTtcXV3Zv38/e/fupXDhwpQpU+a5rlepUiX1cwsLC6ytrdVsn8jISGrXro2i/G80np6eJCQkcO3atRe6vxfxNk4fzCssTQ1Z0LU62/zrs6BrdXlGQgghxEskQTIhhBAiD3vw4AE+Pj5YWFjg6OjIjBkz9KYwJicnM2TIEIoUKYKFhQU1a9YkNDRUPT84OBhbW1u2bduGq6srlpaWeHt7q9P9Mi1cuBBXV1dMTU0pV66c3vS/K1euoCgKq1atokGDBpiamrJs2TJu375Nx44dKVKkCObm5lSsWJEVK1bk+t66d+9Oq1at9PalpqZSsGBBfvnll+d/WG+IgUZhTGu3LIX7AUyLVyLpWgSdy+owMjKiXLlyeHl5ERoayt69e7NkkeWGkZGR3raiKLnOEgPQaDTodPqjfRVBz7dt+qAQQgghhATJhBBCiDxs0KBBhIWFsWHDBnbs2MG+ffs4ceKEerxfv34cPHiQlStXcvr0aT799FO8vb25ePGi2iYxMZFp06axZMkS/vzzT2JiYhgyZIh6fNmyZYwePZoJEyYQGRnJxIkTGTVqFIsWLdIby/DhwxkwYACRkZE0a9aMpKQkqlatyubNmzl79iy9evWic+fO6tS/Z/Hz8yMkJEQvYLdp0yYSExPp0KHDiz6yt4pJsfLoUh6xdskCNSCWGSQLDQ3Fy8sr2/OMjY1JT09/7uu5urpy8OBBvSBYWFgYVlZWFC1aFMiYTvn4M79//z7R0dF6/RgZGb3Q9Z8k0weFEEII8TaR/GwhhBAij0nX6jgSfYerN/4leNEili1dRuPGjYGMYuqFCxcGICYmhqCgIGJiYtR9Q4YMISQkhKCgICZOnAhkZAn9/PPPlC5dGsgIrI0bN0693pgxYwgMDKRdu3YAlCxZkoiICObNm0fXrl3Vdv7+/mqbTI8H27766iu2bdvGb7/9Ro0aNZ55n3Xq1MHFxYUlS5bw9ddfq/f36aefYmlp+XwP7Q1K1+oI2BiR7TEDU0uMCzixa9NqfvxhDgD169enffv2pKam5phJ5uTkRHR0NOHh4RQtWhQrKytMTEyeOZa+ffsyc+ZMvvrqK/r160dUVBRjxoxh0KBBaDQZ7502atSI4OBgWrduja2tLaNHj8bAwCDL9Xft2oWnpycmJibky5fveR6Jnszpg0IIIYQQb5oEyYQQQog8JORsLAEbI4iNTyIl7i/SUlOZfDwVa7dYvCs4YmNjg4uLCwBnzpwhPT2dsmXL6vWRnJyMvb29um1ubq4GyCBjdcTMGlYPHz7k8uXL9OjRg549e6pt0tLSsLGx0eu3WrVqetvp6elMnDiR3377jevXr5OSkkJycjLm5ua5vl8/Pz/mz5/P119/zc2bN9m6dSu7d+/O9flvgyPRd4iNz36FS8ioS5YS9xf5nDOK1NvZ2eHm5sbNmzfVr+WTPv74Y9asWUPDhg25d+8eQUFB+Pr6PnMsRYoUYcuWLQwdOpTKlStjZ2dHjx49GDlypNrmm2++ITo6mlatWmFjY8P48eOzZJIFBgYyaNAgFixYQJEiRbhy5cqzH4QQQgghxFtOgmRCCCFEHhFyNpY+S09kqW1160EyfZaeYG4nD7wrOKr7ExISMDAw4Pjx41kygR7PxMquhlXmdLyEhAQAFixYQM2aNfXaPdmnhYWF3vb333/PrFmzmDlzJhUrVsTCwgJ/f39SUlJyfc9dunRh+PDhHDx4kAMHDlCyZEnq1ctbU/HiHuQcIAOwa9ILuya9MC9YTN0XHh6u18bX11cvCGZiYsIff/yRpa8na4kB3Lt3T2+7QYMGT53yam1tzcqVK/X2PZ4xCNC6dWtat26dYx9CCCGEEHmR1CQTQggh8oDMKXuPh0AMbRxAY0hSbEZ9sYCNEdy5e48LFy4AUKVKFdLT04mLi8PZ2Vnvw8HBIVfXLVSoEIULF+avv/7K0kfJkiWfem5YWBht2rShU6dOVK5cmVKlSqljyy17e3vatm1LUFAQwcHBdOvW7bnOfxsUtDJ9qe2E+C8eX9gDMqbOzpw585Vec+zYsbi7u7/Sa7yvMhdfedkURWHdunUvvV8hhHjbSZBMCCGEyAOym7KnMTHHskIj7u35lUdXT3P1UhQfd+yCRqNBURTKli2Lj48PXbp0Yc2aNURHR3PkyBEmTZrE5s2bc33tgIAAJk2axOzZs7lw4QJnzpwhKCiI6dOnP/W8MmXKsGPHDg4cOEBkZCS9e/fm5s2bz33vfn5+LFq0iMjIyCwZTXlBjZJ2ONqYouRwXAEcbUypUdLudQ5LCACOHj1Kr169Xlp/2QVXhgwZwq5du17aNd5X2QU0O3To8NxvPuRGbGwszZs3B/63gvGTGa5CCPEukiCZEEIIkQfkNGUvXyM/jIuUI251AHGrRlKqggeurq6YmmZkJQUFBdGlSxcGDx6Mi4sLbdu25ejRoxQvXjzX1/bz82PhwoUEBQVRsWJFGjRoQHBw8DMzyUaOHImHhwfNmjXDy8sLBwcH2rZtm+vrZmrSpAmOjo40a9ZMXYAgLzHQKIxp7QaQJVCWuT2mtRsGmpzCaEI8W2pq6gudV6BAgeeqE/giLC0t9eogipfHzMyMggULvvR+HRwccrUYiBBCvGskSCaEEELkATlNxdOYmFOg9VCKD1pN0X5L8OnanaioKJydnYGMemMBAQFER0eTkpLCP//8w5o1a6hYsSKQUevqyZpVbdu2zVLb6vPPP+fkyZMkJydz584d9u7dy0cffQRkZDfodLos06ns7OxYt24dDx484ObNm4wfP55FixbpZZmEhobqZUZcuXJFbyoYZCwecPfuXXr06JHLp/X28a7gyNxOHjjY6H8dHWxMs9SSE2+3Bw8e4OPjg4WFBY6OjsyYMUNvCmNycjJDhgyhSJEiWFhYULNmTUJDQ9XzM6fHbdu2DVdXVywtLfH29iY2NlbvOgsXLlQD3uXKleOnn35Sj2Vm9qxatYoGDRpgamrKsmXLuH37Nh07dqRIkSKYm5tTsWJFVqxY8dT7eTw7KTg4GEVRsnyMHTsWyMg6a9q0Kfnz58fGxoYGDRpw4sQJvb4APvroIxRFUbefnG6p1WoZN24cRYsWxcTEBHd3d0JCQrLcX+biFObm5lSuXJmDBw/m4iv05mi1WqZOnYqzszMmJiYUL16cCRMmABkLqTRq1AgzMzPs7e3p1auXWvMRMn4Xt23blmnTpuHo6Ii9vT1ffvmlGvz08vLi6tWrDBw4UP26QNbplpnPesmSJTg5OWFjY8Nnn33GgwcP1DbZZaS5u7urX2fQzwjMfEOkSpUqKIqCl5cXf/75J0ZGRty4cUOvH39//zxXN1IIIR4nQTIhhBCCrHV63jaZU/aSYk5zdUortEkZL67u7V/O1cCPSbsbi3XC38wa2R+ANm3avMnhvhRarZa4uDjGjx+Pra0tH3744Zse0n/iXcGR/cMasaJnLWZ95s6KnrXYP6yRBMjymEGDBhEWFsaGDRvYsWMH+/bt0wsU9evXj4MHD7Jy5UpOnz7Np59+ire3NxcvXlTbJCYmMm3aNJYsWcKff/5JTEwMQ4YMUY8vW7aM0aNHM2HCBCIjI5k4cSKjRo1i0aJFemMZPnw4AwYMIDIykmbNmpGUlETVqlXZvHkzZ8+epVevXnTu3PmpCzU8rkOHDsTGxqofK1aswNDQEE9PTyAjQNi1a1f279/PoUOHKFOmDC1atFADMEePHgUyMlhjY2PV7SfNmjWLwMBApk2bxunTp2nWrBkffvih3jMCGDFiBEOGDCE8PJyyZcvSsWNH0tLScnUvb8I333zD5MmTGTVqFBERESxfvpxChQrx8OFDmjVrRr58+Th69Ci///47O3fupF+/fnrn79mzh8uXL7Nnzx4WLVpEcHAwwcHBAKxZs4aiRYsybtw49euTk8uXL7Nu3To2bdrEpk2b2Lt3L5MnT37h+8r8/tm5cyexsbGsWbOG+vXrU6pUKZYsWaK2S01NZdmyZXTv3v2FryWEEG+c7h0THx+vA3Tx8fFveihCCCHykAYNGugGDBjwpofxVFvP/KMr1HGiDtAVG7BSV2LYJl2hTt/rjAo46RQjU52lta2uSZMmutOnT7/pob4U0dHROkBXtGhR3c6dO9/0cITQ3b9/X2dkZKT7/fff1X337t3TmZub6wYMGKC7evWqzsDAQHf9+nW98xo3bqz75ptvdDqdThcUFKQDdJcuXVKP//jjj7pChQqp26VLl9YtX75cr4/x48frateurdPp/vezMXPmzGeOuWXLlrrBgwer20/+ritRooRuxowZWc67dOmSzs7OTjd16tQc+05PT9dZWVnpNm7cqO4DdGvXrtVrN2bMGF3lypXV7cKFC+smTJig16Z69eq6vn376t3fwoUL1ePnzp3TAbrIyMin3e5rlZau1R249K9u3clruh3h0ToTExPdggULsrSbP3++Ll++fLqEhAR13+bNm3UajUZ348YNnU6n03Xt2lVXokQJXVpamtrm008/1XXo0EHdzu5rFRQUpLOxsVG3x4wZozM3N9fdv39f3Td06FBdzZo1n9pP5cqVdWPGjFG3H/86Zn49Tp48qXfOlClTdK6urur26tWrdZaWlnr3KYQQb4vcxooM31BsTgghhBDPybuCI4OalmXYY7OnTIu4UnXgQsa0dnvnMpIyp3EK8Sala3Ucib5D3IMk4q9dIjU1lRo1aqjHbWxscHFxATKm1KWnp1O2bFm9PpKTk/Vqcpmbm1O6dGl129HRkbi4OCBjevHly5fp0aMHPXv2VNukpaVhY2Oj12+1atX0x5qezsSJE/ntt9+4fv06KSkpJCcnP3fNsfj4eFq1akXLli0ZOnSouv/mzZuMHDmS0NBQ4uLiSE9PJzExkZiYmFz3ff/+ff755x81Oy2Tp6cnp06d0ttXqVIl9XNHx4zfb3FxcZQrV+657udVCDkbS8DGCHVBleR/okhOTsawaMUsbSMjI6lcuTIWFhbqPk9PT7RaLVFRURQqVAiA8uXLY2BgoLZxdHTkzJkzzz02JycnrKys9PrJ/P56mXx9fRk5ciSHDh2iVq1aBAcH0759e737FEKIvEamWwohhHjvPHz4kC5dumBpaYmjoyOBgYF6x3NbU2jdunWUKVMGU1NTmjVrxt9//63Xz/r16/Hw8MDU1JRSpUoREBCgN1VIURQWLlzIRx99hLm5OWXKlGHDhg16fWzZsoWyZctiZmZGw4YNKahkTGta2LU6sz5zp5NdNJFTPlYDZLmpR/OsmkpCiAwhZ2OpO2U3HRccYsDKcIavOQ1AaFT2q7QmJCRgYGDA8ePHCQ8PVz8iIyOZNWuW2s7IyEjvPEVR1IBwZp2qBQsW6PVx9uxZDh06pHfek8GI77//nlmzZjFs2DD27NlDeHg4zZo1IyUlJdf3nJ6eTocOHbC2tmb+/Pl6x7p27Up4eDizZs3iwIEDhIeHY29v/1z9P4/Hn1NmDS6tVvtKrvU8Qs7G0mfpCb0VhxWjjCL3I9edJeRszlMhnya774sXud9n9aPRaLK8AfEiCz8ULFiQ1q1bExQUxM2bN9m6datMtRRC5HkSJBNCCPHeGTp0KHv37mX9+vVs376d0NDQF6opNGHCBBYvXkxYWBj37t3js88+U4/v27ePLl26MGDAACIiIpg3bx7BwcFqEedMAQEBtG/fntOnT9OiRQt8fHy4c+cOAH///Tft2rWjdevWhIeH4+fnx/Dhw4GMGmVt3IvgXNAyy/09qx7Ns2oqCSGyD4QY2jiAxpAhP61VAyHx8fFcuHAByChsnp6eTlxcHM7OznofDg4OubpuoUKFKFy4MH/99VeWPp61omxYWBht2rShU6dOVK5cmVKlSqljy62BAwdy5swZ1q1bp66S+3j//fv3p0WLFpQvXx4TExP+/fdfvTZGRkakp6fn2L+1tTWFCxcmLCwsS99ubm7PNdY3IV2rI2BjBE/muBrlK4xiaELS1VMEbIwgXfu/Fq6urpw6dYqHDx+q+8LCwtBoNGoWYm4YGxs/9dnmVoECBfRqmt2/f5/o6OinXhfI9tp+fn6sWrWK+fPnU7p06SwZgkIIkddIkEwIIcQ7L12r4+Dl26wPv86u01f55ZdfmDZtGo0bN6ZixYosWrRIzfCKiYkhKCiI33//nXr16lG6dGmGDBlC3bp1CQoKUvtMTU1lzpw51K5dm6pVq7Jo0SIOHDigFjgOCAhg+PDhdO3alVKlStG0aVPGjx/PvHnz9Mbm6+tLx44dcXZ2ZuLEiSQkJKh9zJ07l9KlSxMYGIiLiws+Pj74+vo+8361Wi3BwcFUqFCBevXq0blzZ3bt2gVkZJEtWrRIvf8KFSoQFBT0Ul54CfGuyCkQojExx7JCI+7u+ZVBM5dz+sxZevTogUajQVEUypYti4+PD126dGHNmjVER0dz5MgRJk2axObNm3N9/YCAACZNmsTs2bO5cOECZ86cISgoiOnTpz/1vDJlyrBjxw4OHDhAZGQkvXv35ubN7LPeshMUFMRPP/3Ezz//jKIo3Lhxgxs3bqjZbWXKlGHJkiVERkZy+PBhfHx8MDMz0+vDycmJXbt2cePGDe7evZvtdYYOHcqUKVNYtWoVUVFRDB8+nPDwcAYMGJDrsb4pR6Lv6AVOMymGxljX/Ji7oUFcDNvMmtDjHDp0iF9++QUfHx9MTU3p2rUrZ8+eZc+ePXz11Vd07txZnWqZG05OTvz5559cv349S3DyeTRq1IglS5awb98+zpw5Q9euXfWmeT6pYMGCmJmZERISws2bN4mPj1ePNWvWDGtra7777ju6dev2wmMSQoi3hQTJhBCvXHbLyT9em0WIV+nJ6VKdZ24gJSWFR7ZOahs7O7tsawpZWlqqH3v37uXy5cvqOYaGhlSvXl3dLleuHLa2tkRGRgJw6tQpxo0bp9dHz549iY2NJTExUT3v8Zo7FhYWWFtbq7VjIiMjqVmzpt791K5d+5n3/LR6NH/99ddTayoJIXIOhADka+SHcZFynF88gkaNG+Pp6Ymrq6uadRUUFESXLl0YPHgwLi4utG3blqNHj1K8ePFcX9/Pz4+FCxcSFBRExYoVadCgAcHBwc/MJBs5ciQeHh40a9YMLy8vHBwcaNu2ba6vu3fvXtLT0/nwww9xdHRUP6ZNmwbAL7/8wt27d/Hw8KBz587079+fggUL6vURGBjIjh07KFasGFWqVMn2Ov3792fQoEEMHjyYihUrEhISwoYNGyhTpkyux/qmxD3I/vsCwMbzM6yrf8S9fcv4vFkdOnToQFxcHObm5mzbto07d+5QvXp1PvnkExo3bsycOXOe69rjxo3jypUrlC5dmgIFCrzwPXzzzTc0aNBArTvXtm1bvRp5TzI0NGT27NnMmzePwoUL662erNFo8PX1JT09nS5durzwmIQQ4m2h6N6xirj379/HxsaG+Ph4rK2t3/RwhBD8r46Ira0tDx8+VOteXL58mVKlSr3JoYl3XOZ0qcf/o0uJ+4vYoP4U7fMrC/p6q7W8qlSpQoMGDahduzY+Pj6cO3cuyzvrlpaWODg4EBwcTM+ePUlOTkaj+d/7Tfny5WPmzJl07doVMzMzAgICaNeuXZZxlSpVSs08Wbt2rd6LWFtbW2bOnImvry8fffQR+fLl49dff1WPr1+/nrZt23L37l1sbW0JDg7G39+fe/fuARk1ydatW0d4eLh6zsyZM5k5cyZXrlzh1KlTuLu7c/XqVb0X7R4eHtSvX5+ZM2c+/4MW4h2zPvw6A1aGP7PdrM/caVLGliJFihAYGEiPHj1e/eDEG3Xw8m06Ljj0zHYretaidun34w3BHj16cOvWrSw1NYUQ4m2S21iRrG4phHjpjv91l4/nH1C3j12+Q9VS+dRtT09PDhw4gJubG0lJOb8jK8R/kdN0KUNbR9AYkvzPBQI2FqepmwP34+9x4cIFGjRooFdTqF69ejn2n5aWxrFjx9SMrKioKO7du4erqyuQEXSKiorC2dn5he/B1dU1y4uOJwt3P69SpUphZGSkl9mSWVOpfv36/6lvId4VBa1MczyWcvMyqbevYexYlrtXzfAZOxtAL7tGvLtqlLTD0caUG/FJWf5/AVAABxtTapS0e91De+3i4+M5c+YMy5cvlwCZEOKdIUEyIcRL5TQ8a82VauWKQOqjLPuTk5Nfx5DEeyqn6VIaYzMsKzXlzp5fUcysWLHNhN9//l7NCHu8plBgYCBVqlTh1q1b7Nq1i0qVKtGyZUsgozj1V199xezZszE0NKRfv37UqlVLDZqNHj2aVq1aUbx4cT755BM0Gg2nTp3i7NmzfPfdd7m6hy+++ILAwECGDh2Kn58fx48fJzg4+D89FysrK7p27crQoUOxs7OjYMGCjBkzRs1sE0I8OxBy/8ga0u5eZ9DvZlStWpV9+/aRP3/+1z5O8foZaBTGtHajz9ITKKD3/ZH5G3RMazcMNO/+79M2bdpw5MgRvvjiC5o2bfqmhyOEEC+F1CQTQrw02QXIAL0Ambm5eZYiv0K8Ck+rG5OvYXdMi5Xn1upxfNW5HXXr1qVq1arq8dzUFDI3N2fYsGF8/vnneHp6YmlpyapVq9TjzZo1Y9OmTWzfvp3q1atTq1YtZsyYQYkSJXJ9D8WLF2f16tWsW7eOypUr8/PPPzNx4sTnfBJZTZ8+ndq1a9OqVSuaNGmSpaaSEO+7zEAI/C/wkcmkUGkK+85i8/G/uHPnDjt27KBixYqvf5DijfGu4MjcTh442Oj/znSwMWVuJw91Gv+7LjQ0lMTERGbMmPGmhyKEEC+N1CQTQrwUT06xzHR92XDSrp0FoMSwTazuVYfqzvZk/up5x34FibfIq6wb82QdsLzu4cOHUlNJiGyEnI0lYGOEXlaqo40pY1q7vTeBEJGzdK2OI9F3iHuQREGrjCmW70MGmRBC5EVSk0wI8VplFyADMG/Ri/vz+wNwdUZ7qk1JzLadEC+b1I3J2cmTJzl//jw1atQgPj6ecePGAVJTSYgneVdwpKmbgwRCRLYMNMp7U5xfCCHeFxIkE0K8UvnyleJ+5kaKBMjE6yN1Y55u2rRpREVFYWxsLDWVhHgKCYQIIYQQ7w+ZbimEeClyrEeWjSuTW77CkQihT6ZLifeVn58fv/zyyzOntSuKwvDhw5k0adJrGpkQQgghxOsl0y2FEK/V6l51cpxy+WQ7IV4nmS4l3ldTp05l4MCB6raXlxeHDx/m0aOsqw0LIYQQQggJkgkhXpKqpfK91HZCvEwyXUq8j+zs7LCze/9q7gkhhBBCvCjNmx6AEOLd8axplDLNUggh/pvRo0ejKAopKSkArFq1CkVRqFWrltrGxcWFkiVL4ufnh6JkZEz6+fmxd+9ekpKSUBQFRVHw8/NTz/nnn38oXLgwiqJgbGzMiBEjXu+NCSGEEEK8BSRIJoR4qa5MbpllSuXqXnUkQCaEEC9BZmBr5cqVwP+CZGfPnlXbREdHU79+fb3zpk6dStWqVTExMeHUqVOcOnWKqVOnqseXLVvGxx9/zM6dO3Fzc2PixIlcvnz5NdyREEIIIcTbQ6ZbCiFeuqql8klQTAghXpKUNC1LDl7h6p1EStiZY2Zmxm+//UaXLl04dOgQTZo0YceOHdy4cYNr166RmppKp06dWLVqldqHnZ0dlpaWKIpCpUqVslyjdu3a/PDDDwBs376dQoUKsWLFCkaOHPna7lMIIYQQ4k2TIJkQQgghxFtq0pYIFuyLRvvYApXp1oUJ3X8QrVbLjRs3WLhwIfv27WPhwoXExsai0Who2rSpXpDsWapXr65+XrBgQQBiYmJe2n0IIYQQQuQFMt1SCCGEeAonJydmzpypbiuKwrp16556jq+vL23btn2l4xLvvklbIpj3p36ADMC0ZFUext/h82+moygKLVq0oFy5cmzatIndu3dTrFix576WiYlJln1arfZFhy6EEEIIkSdJkEwIIYR4DrGxsTRv3hyAK1euoCgK4eHhem1mzZpFcHDw6x+ceGekpGlZsC8622OWlT8AYPX8QIoUKQKAt7c3Z8+e5fLly9StWzfb84yNjdHpdNkeEyIvCgkJoW7dutja2mJvb0+rVq3UWnqZv5/XrFlDw4YNMTc3p3Llyhw8eBCAhw8fYm1tzR9//KHX57p167CwsODBgwev/X6EEEK8eRIkE0IIIZ6Dg4NDtlk3j7OxscHW1vb1DEi8k5YcvJIlgyyTka0DGBqTdu8Gji7uQEZB/4cPH5KamoqPj0+255UpU4bk5GRWrVpFVFQU9+/ff0WjfzWsrKxQFIXx48e/6aGIt8TDhw8ZNGgQx44dY9euXWg0Gj766CO9LMgRI0YwZMgQwsPDKVu2LB07diQtLQ0LCws+++wzgoKC9PoMCgrik08+wcrK6nXfjhBCiLeABMmEEELkGVqtlqlTp+Ls7IyJiQnFixdnwoQJAJw5c4ZGjRphZmaGvb09vXr1IiEhQT03cwrktGnTcHR0xN7eni+//JLU1FS1TVxcHK1bt8bMzIySJUuybNmyLGN4fLplyZIlAahSpQqKouDl5aV3rUzJycn079+fggULYmpqSt26dTl69Kh6PDQ0FEVR2LVrF9WqVcPc3Jw6deoQFRWltjl16hQNGzbEysoKa2trqlatyrFjx/7zMxVvp6t3Ep963MiuKAClankDULp0aUxNTdFoNGqm45MmTZpEkSJF6NixI+XKlWPQoEEvd9AvmaIoKIrypoch3iLpWh0HL99mffh1Dl6+TduP2tGuXTucnZ1xd3fn119/5cyZM0RERKjnDBkyhJYtW1K2bFkCAgK4evUqly5dAjKCy9u2bSM2NhbI+D9gy5YtdO/e/Y3cnxBCiDdPCvcLIYTIM7755hsWLFjAjBkzqFu3LrGxsZw/f56HDx/SrFkzateuzdGjR4mLi8PPz49+/frpTXvcs2cPjo6O7Nmzh0uXLtGhQwfc3d3p2bMnkBHc+ueff9izZw9GRkb079+fuLi4HMdz5MgRatSowc6dOylfvjzGxsbZtvv6669ZvXo1ixYtokSJEkydOpVmzZpx6dIl7Ozs1HYjRowgMDCQAgUK8MUXX9C9e3fCwsIA8PHxoUqVKsydOxcDAwPCw8MxMjJ6CU9VvI1K2Jk/9XjhbrMBaPKBq7rv0aNHem0WLlzIwoUL1W1ra2uuXbuWpa/spmDKtEzxtgk5G0vAxghi45PUfbap/2IbuZaY86f5999/1QyymJgY3NzcAPRWc3V0dAQygmHlypWjRo0alC9fnkWLFjF8+HCWLl1KiRIlqF+//mu8MyGEEG8TySQTQgjxVsvMHFgZFsXMmbOYPHkKXbt2pXTp0tStWxc/Pz+WL19OUlISixcvpkKFCjRq1Ig5c+awZMkSbt68qfaVL18+5syZQ7ly5WjVqhUtW7Zk165dAFy4cIGtW7eyYMECatWqRdWqVfnll1+yBB4eV6BAAQDs7e1xcHDQC3hlevjwIXPnzuX777+nefPmuLm5sWDBAszMzPjll1/02k6YMIEGDRrg5ubG8OHDOXDgAElJGS8IY2JiaNKkCeXKlaNMmTJ8+umnVK5c+T8/X/F26lzbCc0zkqg0Ska7d9HjGWSZGWUPHz4EYPLkyeo+RVGy/BwYGRnpHVcUhfPnz7/W8YuXK+RsLH2WntALkAGcCx7J4cgYen47mcOHD3P48GEAUlJS1DaPv5mQ+X31+HRMPz8/9c2UoKAgunXrJhmMQgjxHpMgmRBCiLdWyNlY6k7ZTccFhxg4bzMpKcnMvWhGyNlYvXaRkZFUrlwZCwsLdZ+npydarVZvymL58uUxMDBQtx0dHdVMscjISAwNDalatap6vFy5cv+5ttjly5dJTU3F09NT3WdkZESNGjWIjIzUa5tTxgPAoEGD8PPzo0mTJkyePFktTi3eTcaGGnrWK/nUNj3rlcTY8N38U27SpEnq5506daJTp06YmZkBkJiYiJWVFeXKlQPg9OnTalsbGxvS0tJQFIU6deqo57i6uiLypnStjoCNETyZ25j+6D5pd65hW6cDG27ZU9alHHfv3n3u/jt16sTVq1eZPXs2ERERdO3a9eUMXAghRJ70bv5lJYQQIs97MnNAMcooln/rQTJ9lp7IEijLjSenJyqKopdR8KY9LeNh7NixnDt3jpYtW7J7927c3NxYu3btGxmneD2+aeFG7/ols2SUaRToXb8k37RwezMDe4Vi/k3EbdRW5t2rqO5bsmQJS5YsQaPJ+LPV3Nyc+/fvExkZqS6isXXrVgB1MQKtVktYWBiJiYmSFZTHHYm+kyWDDEBjaonGzJoHp7YRc+Uvflq+7oXq7OXLl4927doxdOhQPvjgA4oWLfoyhi2EECKPkiCZEEKIt052mQNG+QqjGJrw6OopAAI2RpD+/8v/ubq6curUKXU6FkBYWBgajQYXF5dcXbNcuXKkpaVx/PhxdV9UVBT37t3L8ZzMGmTp6ek5tildujTGxsZqbTGA1NRUjh49qtbMya2yZcsycOBAtm/fTrt27bKsyva28/Lywt/fHwAnJydmzpz5RseTF3zTwo3z45szqqUrXWqXYFRLV86Pb/5OBsicv91M/Wl7SEzV6v3sO3+7Wa9dly5d1M8LFiwIwKxZs/TaPD7VMrO+2sqVK1/NwMUrFfcga4AMQFE05P/wa1JuXOKfX75k+rgRfP/99y90jR49epCSkiIF+4UQQkjhfiGEeFW8vLxwd3eXQMALyC5zQDE0xrrmx9wLDUIxMCSliBu/rtmOJv4aPj4+jBkzhq5duzJ27Fhu3brFV199RefOnSlUqFCuruni4oK3tze9e/dm7ty5GBoa4u/vr07Xyk7BggUxMzMjJCSEokWLYmpqio2NjV4bCwsL+vTpw9ChQ7Gzs6N48eJMnTqVxMREevTokauxPXr0iKFDh/LJJ59QsmRJrl27xtGjR/n4449zdf7b6OjRo3rTY0XOjA019KhX6k0P45Vy/nYzaTkkdaZp9QNlhQsXVj/PzBLLrN2XqXr16ln6adKkyUsYqXjdClqZ5njMzMkdM7+5ACzvWYvape31Fp14cgEKW1vbbBeluH79Ovb29rRp0+YljVoIIUReJUEyIYR4SwUHB+Pv7//UTKZ3VU6ZAzaen6FoDLi3bxnpCXcYsaEQA/t/ibm5Odu2bWPAgAFUr14dc3NzPv74Y6ZPn/5c1w0KCsLPz48GDRpQqFAhvvvuO0aNGpVje0NDQ2bPns24ceMYPXo09erVIzQ0NEu7yZMno9Vq6dy5Mw8ePKBatWps27aNfPny5WpcBgYG3L59my5dunDz5k3y589Pu3btCAgIeK77e5tkLnogRMy/iTkGyDKlaUH7HCtuHjly5D+OSrwtapS0w9HGlBvxSVnqkgEogIONKTVKZl045VkSExOJjY1l8uTJ9O7dO8cVioUQQrw/JEgmhBDirZNT5oCiaLCp0wGbOh0AWPH/mQMAFStWZPfu3Tn2mbl62eOezPJzcHBg06ZNevs6d+6st/1kFoKfnx9+fn5PvZapqSmzZ89m9uzZ2Y7Ny8srS7/u7u56+1asWJHtuXmVk5MT/v7+6vRLRVFYsGABmzdvZtu2bRQpUoTAwEA+/PBD9ZyzZ88ydOhQ9u3bh4WFBR988AEzZswgf/78b+guxMvgPWvvU4//s2w4RvmdSEzJeVpzJlNTU5KSklAUBRcXFwwMDDh//jxarTbbDCLx9jPQKIxp7UafpSdQQC9QllltbkxrNwyetRxsNqZOncqECROoX78+33zzzcsYrhBCiDxOapIJIcQrlJaWRr9+/bCxsSF//vyMGjVKfaGWnJzMkCFDKFKkCBYWFtSsWVPNQgoNDaVbt27Ex8erdXXGjh3LnDlzqFChgtr/unXrUBSFn3/+Wd3XpEkTRo4cqW6vX78eDw8PTE1NKVWqFAEBAaSlpanH7927h5+fHwUKFMDa2ppGjRpx6tQp9fjYsWNxd3dnyZIlODk5YWNjw2effcaDBw9e1WNTMwdyesmjAI4vmDkg3l4BAQG0b9+e06dP06JFC3x8fLhz5w6Q8X3aqFEjqlSpwrFjxwgJCeHmzZu0b9/+DY9a/FePUrNPIzMskDHFNPXaWRLDN0Fq9hmmen09eoSpaUaQPSoqioiIiLdqcQ7xYrwrODK3kwcONvpvoDjYmDK3kwfeFRxfqN+xY8eSmprKrl27sLS0fBlDFUIIkcdJJpkQQrxCixYtokePHhw5coRjx47Rq1cvihcvTs+ePenXrx8RERGsXLmSwoULs3btWry9vTlz5gx16tRh5syZjB49mqioKAAsLS2Jjo6mf//+3Lp1iwIFCrB3717y589PaGgoX3zxBampqRw8eJDhw4cDsG/fPrp06cLs2bOpV68ely9fplevXgCMGTMGgE8//RQzMzO2bt2KjY0N8+bNo3Hjxly4cAE7u4wg1OXLl1m3bh2bNm3i7t27tG/fnsmTJzNhwoRX8txeZeaAeH3StTqORN8h7kESBa1Ms50q9ThfX186duwIwMSJE5k9ezZHjhzB29ubOXPmUKVKFSZOnKi2//XXXylWrBgXLlygbNmyr/BOxKtkZqQhMZtAWZHu+pmX5kYaIsY319t39erVLOc9evTo5Q5QvBW8KzjS1M1B73dKjZJ28v+AEEKIl0qCZEII8RI9HhS4/yiVYsWKMWPGDHXqz5kzZ5gxYwbNmjUjKCiImJgYtQj1kCFDCAkJISgoiIkTJ2JjY4OiKDg4OKj9V6hQATs7O/bu3csnn3xCaGgogwcPVld2O3LkCKmpqdSpUwfIyMwZPnw4Xbt2BaBUqVKMHz+er7/+mjFjxrB//36OHDlCXFwcJiYmAEybNo1169bxxx9/qAE1rVZLcHAwVlZWQMYUxF27dr2yIBn8L3MgYGOEXhF/BxtTxrR2e+HMAfF6hJyNzfK1uxNzl3zFHuZ4TqVKldTPLSwssLa2Ji4uDoBTp06xZ8+ebLM9Ll++LEGyPCxkQAPqT9uTq3bi/WagUdQp9kIIIcSrIEEyIYR4SZ4MCtyIvY91weJsO3dDDejUrl2bwMBAzpw5Q3p6epYX9snJydjb5/wCQFEU6tevT2hoKE2aNCEiIoK+ffsydepUzp8/z969e9XC9ZARWAgLC9MLZqWnp5OUlERiYiKnTp0iISEhyzUfPXrE5cuX1W0nJyc1QAbg6OioBi9eJckcyJtCzsbSZ+mJLJljKWladkfGEXI2NtvzjIyM9LYVRVGnyiUkJNC6dWumTJmS5TxHRwmY5mXF85tjqOGpxfsNNRnthBBCCCFeJQmSCSHES5BTUOBRSjp9lp7IUjMlISEBAwMDjh8/joGBgd45z6qL4uXlxfz589m3bx9VqlTB2tpaDZzt3buXBg3+l22RkJBAQEAA7dq1y9KPqakpCQkJODo6Zrsio62trfr504IXr5pkDuQt6VodARsjnjq1MmBjxHP36+HhwerVq3FycsLQ8P3888XQ0JDWrVuzdu3aNz2Ul+7SxJY4f7s520CZoSbjuBBCCCHEq/Z+/pUphBAv0dOCAsn/XAAyggJN3Rw4dOgQZcqUoUqVKqSnpxMXF0e9evWy7dfY2Jj09KyruTVo0AB/f39+//13vLy8gIzA2c6dOwkLC2Pw4MFqWw8PD6KionB2ds72Gh4eHty4cQNDQ0OcnJye676FyM6R6Dt6UyyzExufRPLT0oay8eWXX7JgwQI6duzI119/jZ2dHZcuXWLlypUsXLgwS7BZ5D2XJrYk5t9EvGft5VGqFjMjDSEDGkgGmRBCCCFeG1ndUggh/qOnBQXSHtzi9q4FxPx1ie9mL+CHH35gwIABlC1bFh8fH7p06cKaNWuIjo7myJEjTJo0ic2bNwMZUxwTEhLYtWsX//77L4mJiUBG3aZ8+fKxfPlyvSDZunXrSE5OxtPTU73+6NGjWbx4MQEBAZw7d47IyEhWrlyprn7ZpEkTateuTdu2bdm+fTtXrlzhwIEDjBgxgmPHjr3CpybeVXEPnr0CIWQEl59H4cKFCQsLIz09nQ8++ICKFSvi7++Pra0tGs3r/XMmLS2NZs2aYWRkhKIomJmZqcHpmTNnoigKEyZMwNzcHEVRMDIyonPnzowbNw4TExMURcHJyYl///1X7dPW1pZKlSpRqVIlFEVBo9FQr149NWPT1taW9PR0dUVbRVGIi4tDURS9wDjAN998g6Io/PPPP6/vobwkxfObEzG+OdGTWxIxvrkEyIQQQgjxWkkmmRBC/EdPCwpYlG+ELi2F2MWDmGZqxIABA9Ri+EFBQXz33XcMHjyY69evkz9/fmrVqkWrVq0AqFOnDl988QUdOnTg9u3bjBkzhrFjx6IoCvXq1WPz5s3UrVsXyAicWVtb4+LigoWFhXr9Zs2asWnTJsaNG8eUKVMwMjKiXLly+Pn5ARnTJrds2cKIESPo1q0bt27dwsHBgfr161OoUKFX9cjEO6yglWmOxxw+n6x+vn5fuN40Wp0ua9Ds3r17ettlypRhzZo1/32Q/1Hz5s35888/GTt2LJ6enixdupTp06dTvHhxtc3kyZMJDAzE1tYWHx8ffvvtN6ysrAgODub27dv079+fzp07s3XrVvWcM2fO4O7uzpYtW9i4cSNz587F19eXxYsXc/z4cVxcXGjQoAEzZswAoGDBgpQrV44lS5YQGBio9hMcHEypUqXURUGEEEIIIV6El5cX7u7uzJw5800P5bVRdNn9VZqH3b9/HxsbG+Lj47G2tn7TwxFCvAcOXr5NxwWHntluRc9aUltLvPPStTrqTtnNjfikbKcgK2SsULp/WKM8swBDSpqWJQevcPVOIgWM0+nfwp158+apAW8AFxcXkpOT8ff3Z+DAgUydOpWhQ4fi6+vLokWL9Pqzt7fHwMCA9PR0/v33X9q2bcv69f/H3n3H13T/Dxx/nZsdWWIlRhKRQew0KIq0qFi1qqjao6WxarZqBEWNIrSKlqhdtUdjJ0jtxEy+KIm0Gg1CIsi89/dHfjl1JSEIkXg/H488HrnnfM7nvM+p3tzzvp/P+7MFY2NjHj58yD///EO5cuWoXr06ERER3LhxgyFDhrBixQo0Gg3vv/8+/v7+uLq6smzZMnr37k1YWBg1atTgwoULVKlShblz5zJkyJBXfauEEEIIUYgUpiRZbnNFMt1SCCFeUO3ytthbm5LT474C2FtnrMooRGFnoFGY0NoDIMv/E5mvJ7T2KDAJsmk7w6k47ncm74jglyPXmLRkPQCffvqpOu1RURQuXbqkN30yc0TovHnzKFGiBAAxMTHExMTQrVs30tLSePDgATqdjkOHDgFQvHhxNBoNwcHBlClThhYtWpCSkkL37t05efIkGo2GRo0aodPpaNGiBampqfTq1QtTU1PGjBkDwFdffYWhoSGDBg16lbdJCCGEEKJQkCSZEEK8oMKWFBDiRflUsWfhJ57YWetPvbSzNs2y0uvrbNrOcBYdjOTR8mnah/cAKNq4H72/+Ym9e/eqP/v27VPbmZiaceTKbYIiE0nXZfy/b2dnh52dHe+99x7x8fHodDrOnj2LsbGxOooMICgoSG+V2u3bt/PTTz+hKArW1tasWrWK69evs3nzZgBatmzJgQMHANi1axfe3t6vvE6bEEIIIQqntLQ0fH19sba2pnjx4owbNw6dTsekSZOoUqVKlvY1atRg3Lhx+RBp3pCaZEIIkQcykwJ+28L1ivjbWZsyobVHgUkKCJFXfKrY09TDjuORccTeS6KkZcZoyoKSLE5J07LkUGSW7WbObwGQeusvDtxrw8JG72Js+F9C6siRIwB0WBjCHY0NAHcfpgAQeD4Gnyr2NGjQgPT0dAwMDAgODqZRo0Zs27ZNrcEWHBzMyJEjWbp0KYaGGR/V6tSpg0ajIS0tjWLFiuHu7k5ERAQAM2fOZMOGDXTo0IHk5GSmTp36Uu5JQRIVFUX58uXVaahCCCGEeD7Lly+nT58+HD9+nJMnT9K/f38cHBzo3bs3fn5+nDhxglq1agEQFhbG2bNnX4sass9LkmRCCJFHCnpSQIi8ZqBRCmwdvhVHoshuAU5Dy2IYl/Eg8cwudDotX9ndoWZxhU2bNlG0aFE0tmUBuHkvGUNr/WMHrAxVR9IVKVKE5ORkgoODadq0Kb///js6nY7KlStz+fJlgoODOXr0KA0bNuSPP/4AwNLSkpMnT3Ly5EmSkv5LxpcvXx5HR0c2btxI8eLF1Q+qQgghhBDPKl2rU59nEh6mUq5cOebMmYOiKLi7u3Pu3DnmzJlDv379aNasGcuWLVM/eyxbtoxGjRrh7Oycz1fx/GQsvhBC5KHMpECbGmWoW6GYJMiEKKCuxT3IcV+pj6dj5vo298/vZ/bQbnTr1o2goCAqV6lK4Pkb2Rzx38ctv23hpGt12NjYoNVqOXjwIN7e3iiKgomJCdevXwdg9erV1KtXj8WLF5OWlsaxY8f49ttvuXv3LrVq1eJ///sfHh4ear+ZNci6deuWNzdACCGEEG+cwPMxvPPtfrosOcqQtacJj0nglrkDuy789/mmbt26XL58mfT0dPr168eaNWtISkoiJSWF1atX07t373y8ghcnSTIhhBBCiMc42prnuE+j0VCy/VgcR27mp4NX0Gq13Lp1izqtuqJUbYXj6O0YWpdS25u71MLY3pXU+H/5O+Zfjl65xffff4+iKBgaGlKxYkUgY6RYYmIinTt3RqvVEhISgru7O23atKFfv35UrFiRo0eP4uPjg4uLC23atFHPcfnyZRRFYdKkSS/vpryGtFotM2bMwMXFBRMTExwcHPjmm2/U/VevXuXdd9/F3Nyc6tWrq9NhM23YsIHKlStjYmKCk5MTs2fP1tv/ww8/4OrqiqmpKaVKleLDDz/UO/e0adMoX748ZmZmVK9end9++03dHxQUhKIo7Nu3Dy8vL8zNzalXrx4XL158SXdDCCGEeH6B52MYsDJUr3QMwMOUdAasDCXwfEyWY1q3bo2JiQmbNm1i27ZtpKam6v2tLIgkSSaEEEII8ZhudZ142kBQjZLRLlPsvaRs21nVbg+Khn9+Gsjf87ty4fIVGjRogFar1SvQb2lpSXp6Ot7e3nrHL1u2jLfeeotWrVpRt25ddDodO3fuxMjIiFu3brF//34CAgKoW7cuFhYWz3vJBdKXX37J9OnTGTduHOHh4axevZpSpf5LUI4dO5YRI0Zw+vRp3Nzc6NKlC2lpaQCcOnWKjz76iM6dO3Pu3DkmTpzIuHHjCAgIAODkyZMMHjyYSZMmcfHiRQIDA2nYsKHa97Rp0/jll1/48ccfuXDhAsOGDeOTTz4hODhYL8axY8cye/ZsTp48iaGhYYH/hl0IIUThk67V4bctnGwqTZD8zyXgv9HwR48exdXVFQMDAwwNDenRowfLli1j2bJldO7cGTMzs1cbfB5TdDpddvehwEpISMDa2pr4+HisrKzyOxwhhBBCFFCZq1vm5NOG5fmyxX9THo9cuU2XJUef2u+afm9nqdVmY2OjFpp/Ft7e3gQHB1O0aFHCw8Oxs7N7puMLmkfrpBRRUmlVpyILFiygb9++eu0yC/f/9NNP9OnTB4Dw8HAqV65MREQEFStWpGvXrty8eZPdu3erx40aNYodO3Zw4cIFNm7cSK9evfj777+xtLTU6z85ORlbW1v27t1L3bp11e19+/blwYMHrF69mqCgIN5991327t1L48aNAdi5cyctW7bk4cOHmJrqr/4qhBBC5JecPsPcWD2GlH+vYFHtfSxrNKd/ZYWZX3/B7Nmz+fTTT4GM0eyVKlUCICQkhDp16rzS2HMrt7kiKdwvhBBCCJGNzATYkkORekX8NQr0a6CfIAOoXd4We2tTbsQnZftNrELGire1y9tm2Ze5suWzCgoKeq7jCqLA8zF6Kwgn/3OR5ORkDMtWzfGYatWqqb/b22esMhwbG0vFihWJiIjQm7IKUL9+febOnUt6ejpNmzbF0dERZ2dnfHx88PHxoV27dpibm/Pnn3/y4MEDmjZtqnd8SkoKNWvWzFUMDg4Oz3EXhBBCiLyX02h4gCKV30OXlkLML18wy9SIIUOG0L9/f3W/q6sr9erVIy4u7rVNkD0LSZIJIYQQQuTgyxYeDH+/IiuORHEt7gGOtuZ0q+uEsWHWihUGGoUJrT0YsDIUBfQSZZkzNye09pAFPZ5DZp0UvXtqZALA15vPY1fWAZ8q9lmOMzIy+q+9knHftVptrs5paWlJaGgoQUFB7N69m/HjxzNx4kROnDhBYmIiADt27KBMmTJ6x5mYmORZDEIIIcSrcsiiVQAAxhpJREFUUNIy+9HNdh9PV38v1uzzbEfD63Q6/vnnHwYOHPhSY3xVJEkmhBBCCPEExoYa+jTI3VLmPlXsWfiJp96IJ8gYQTahtUe2iRzxZDnVSTEqWhrF0ISka2fw2+ZEUw+7Z0pAVqpUiZCQEL1tISEhuLm5YWBgAIChoSFNmjShSZMmTJgwARsbG/bv30/Tpk0xMTEhOjpar66cEEIIURA972j4mzdvsnbtWm7cuEGvXr1eSawvmyTJhBBCCCHykE8Ve5p62Km1s0paZnyolBFkz+d4ZFyWlbYAFENjrOp04E7QMi4bGLIxyJpy5mlcuHBBrQH2JMOHD6dWrVpMnjyZTp06ceTIERYsWMAPP/wAwPbt27l69SoNGzakaNGi7Ny5E61Wi7u7O5aWlowYMYJhw4ah1Wp55513iI+PJyQkBCsrK3r06JHn90EIIYR4WZ53NHzJkiUpXrw4ixcvpmjRoq8q3JdKkmRCCJENnU5Heno6hobyNimEeHYGGiXLdATxfJ5UJ8W6fmcUjQF3D63i410LKF3ans8++yxX/Xp6evLrr78yfvx4Jk+ejL29PZMmTaJnz55AxmIKGzduZOLEiSQlJeHq6sqaNWuoXLkyAJMnT6ZEiRJMmzaNq1evYmNjg6enJ1999dULX7MQQgjxqj3PaPhCtg4kIKtbCiHeIMnJyYwcOZK1a9eSkJCAl5cXc+bMoVatWuoqZDt37uTrr7/m3Llz7N69G29v7/wOWwgh3mgvsmqoEEIIIZ7NoytJF6bR8LK6pRBCPGbUqFFs2LCB5cuX4+joyIwZM2jWrBl//vmn2mbMmDHMmjULZ2fnQjNkWAghCrIXWTVUCCGEEM/mTR8Nn3VpJiGEKCTStTqOXLnNltPX2X8umoULFzJz5kyaN2+Oh4cHS5YswczMjJ9//lk9ZtKkSTRt2pQKFSpgaysPXEIIkd8y66TAf3VRMuXlqqEpKSkvdLwQQgghCj5JkgkhCqXA8zG88+1+uiw5ypC1p/lkzlZSU1NJLe6qtjEyMqJ27dpERESo27y8vPIjXCGEEI/w9vZm8ODBjBo1CltbW3o2qUn9hP3YWWcsUa9NSuT27/78Pb8r/y7ozIzBXTlz5ox6/JUrV2jTpg2lSpXCwsKCWrVqsXfvXr1zODk5MXnyZLp3746VlRX9+/cnICAAGxsbdu3aRaVKlbCwsMDHx4eYmBi9Y3/66ScqVaqEqakpFStWVIv9Q0ayzdfXF3t7e0xNTXF0dGTatGlARu2WiRMn4uDggImJCaVLl2bw4MEv6zYKIYQQ4hlJkkwIUegEno9hwMrQbFdD+3rTeQLPx2RzVIYiRYq8zNCEEELk0vLlyylSpAjHjh1jxowZrP5xDhM801nT721sj35PjeIKhw/s4UxYKJ6enjRu3Ji4uDgAEhMTadGiBfv27SMsLAwfHx9at25NdHS03jlmzZpF9erVCQsLY9y4cQA8ePCAWbNmsWLFCg4ePEh0dDQjRoxQj1m1ahXjx4/nm2++ISIigqlTpzJu3DiWL18OgL+/P1u3buXXX3/l4sWLrFq1CicnJwA2bNjAnDlzWLRoEZcvX2bz5s1UrVr1FdxNIYQQQuSG1CQTQhQq6VodftvCs9StMbSxBwNDkq6H47fNgaYedmjT0zhx4gRDhw7Nj1CFEEI8QbVq1ZgwYQIArq6uLFiwgKAD+2lVxJzIiDPExsZiYmICZCS7Nm/ezG+//Ub//v2pXr061atXV/uaPHkymzZtYuvWrfj6+qrb33vvPYYPH66+PnToEKmpqfz4449UqFABAF9fXyZNmqS2mTBhArNnz6Z9+/YAlC9fnvDwcBYtWkSPHj2Ijo7G1dWVd955B0VRcHR0VI+Njo7Gzs6OJk2aYGRkhIODA7Vr134Jd08IIYQQz+OVjCT7/vvvcXJywtTUlDp16nD8+PEntl+/fj0VK1bE1NSUqlWrsnPnzlcRphCiEDgeGZftCDKNsSmWNVpw58BSroaFsG73Efr168eDBw/o06dPPkQqhBDiUY/WkUx4mJplhJW9vT2xsbGcOXOGxMREihUrhoWFhfoTGRnJlStXgIyRZCNGjKBSpUrY2NhgYWFBRERElpFk2U2xNzc3VxNkj54X4P79+1y5coU+ffronXvKlCnquXv27Mnp06dxd3dn8ODB7N69W+2rY8eOPHz4EGdnZ/r168emTZtIS0vLmxsohBBCiBf20keSrVu3ji+++IIff/yROnXqMHfuXJo1a8bFixcpWbJklvZ//PEHXbp0Ydq0abRq1YrVq1fTtm1bQkNDqVKlyssOVwhRwMXey5ogy1TUuyeg49b22fTcMpXatbzYtWuXrGIphBD5LPB8DH7bwtUvOW7EJBBz5l8+OB+DTxV7ABRFQavVkpiYiL29PUFBQVn6sbGxAWDEiBHs2bOHWbNm4eLigpmZGR9++GGW4vzZTbE3MjLSe60oCjpdxvjkxMREAJYsWUKdOnX02hkYGADg6elJZGQkv//+O3v37uWjjz6iSZMm/Pbbb5QrV46LFy+yd+9e9uzZw8CBA5k5cybBwcFZziuEEEKIV++ljyT77rvv6NevH7169cLDw4Mff/wRc3Nzli5dmm37efPm4ePjw8iRI6lUqRKTJ0/G09OTBQsWvOxQhRCFQElL0xz3KYbG2Db5lHKDVxMcfp3Dhw9Tq1YtIKNItE6nUx+whBB5y9vbW53a7OTkxNy5c/M1HvH6yKmO5P3kNAasDM1SR9LT05MbN25gaGiIi4uL3k/x4sUBCAkJoWfPnrRr146qVatiZ2dHVFTUC8daqlQpSpcuzdWrV7Ocu3z58mo7KysrOnXqxJIlS1i3bh0bNmxQ66WZmZnRunVr/P39CQoK4siRI5w7d+6FYxNCCCHEi3upI8lSUlI4deoUX375pbpNo9HQpEkTjhw5ku0xR44c4YsvvtDb1qxZMzZv3pxt++TkZJKTk9XXCQkJLx64EKLAql3eFntrU27EJ2WpSwagAHbWptQub/uqQxNC/L8TJ07k6yIZQUFBvPvuu9y5c0cS4/kspzqSj/LbFk5TDzv1dZMmTahbty5t27ZlxowZuLm58c8//7Bjxw7atWuHl5cXrq6ubNy4kdatW6MoCuPGjUOr1eZJzH5+fgwePBhra2t8fHxITk7m5MmT3Llzhy+++ILvvvsOe3t7atasiUajYf369djZ2WFjY0NAQADp6enUqVMHc3NzVq5ciZmZmV7dMiGEEELkn5c6kuzWrVukp6dTqlQpve2lSpXixo0b2R5z48aNZ2o/bdo0rK2t1Z9y5crlTfBCiALJQKMwobUHkJEQe1Tm6wmtPTDQPL5X5IXHpzIJkZ0SJUpgbm6e5/3qdDqp71TA5FRHMpMOiIlP4nhknLpNURR27txJw4YN6dWrF25ubnTu3Jlr166pnyG/++47ihYtSr169WjdujXNmjXD09MzT2Lu27cvP/30E8uWLaNq1ao0atSIgIAAdSSZpaUlM2bMwMvLi1q1ahEVFcXOnTvRaDTY2NiwZMkS6tevT7Vq1di7dy/btm2jWLFieRKbEEIIIV7MKync/zJ9+eWXxMfHqz9//fVXfockhMhnPlXsWfiJJ3bW+lMv7axNWfiJp1rf5k3j7e3NoEGDGDp0KEWLFqVUqVIsWbKE+/fv06tXLywtLXFxceH3339XjwkODqZ27dqYmJhgb2/PmDFj9JIQ3t7e+Pr6MnToUIoXL06zZs0AOH/+PM2bN8fCwoJSpUrRrVs3bt269cqvWeSP+/fv0717dywsLLC3t2f27Nl6+x+dbqnT6Zg4cSIODg6YmJhQunRpBg8erLZNTk5m9OjRlCtXDhMTE1xcXPj555+BjBFhiqLw+++/89Zbb2FiYsLhw4fRarVMmzaN8uXLY2ZmRvXq1fntt98AiIqK4t133wWgaNGiKIpCz549AZ54nHg5cqojaffxdGyb9Ndrt3nzZgICAoCMRJS/vz/Xr18nJSWF6OhoVq5cqX5Z6uTkxP79+3nw4AHR0dF8/vnnBAUF6U3zjYqKyrK6cc+ePbl7967etrZt26o1yTJ9/PHHhIWFkZycTFxcHMHBwbRr1w6Afv36ERYWRmJiIvHx8ezdu5eaNWuqfR09epT4+HgSExM5cuQIjRs3ftbbJoQQQoiX5KUmyYoXL46BgQH//vuv3vZ///0XOzu7bI+xs7N7pvYmJiZYWVnp/QghhE8Vew6Pfo81/d5mXucarOn3NodHv/fGJsgyLV++nOLFi3P8+HEGDRrEgAED6NixI/Xq1SM0NJT333+fbt268eDBA65fv06LFi2oVasWZ86cYeHChfz8889MmTIlS5/GxsaEhITw448/cvfuXd577z1q1qzJyZMnCQwM5N9//+Wjjz7Kp6sWr9rIkSMJDg5my5Yt7N69m6CgIEJDQ7Ntu2HDBubMmcOiRYu4fPkymzdv1lvVsHv37qxZswZ/f38iIiJYtGgRFhYWen2MGTOG6dOnExERQbVq1Zg2bRq//PILP/74IxcuXGDYsGF88sknBAcHU65cOTZs2ADAxYsXiYmJYd68eQBPPC4v/fbbb1StWhUzMzOKFStGkyZNuH//Pj179qRt27b4+flRokQJrKys+Oyzz/RGaAYGBvLOO+9gY2NDsWLFaNWqlbqqYqa///6bLl26YGtrS5EiRfDy8uLYsWPq/i1btuDp6YmpqSnOzs74+fmpye+nJS3z2pPqSD5POyGEEEKIF/FSa5IZGxvz1ltvsW/fPtq2bQtkfEu7b98+fH19sz2mbt267Nu3T++bvT179lC3bt2XGaoQohAy0CjUrfBmT2FJ1+o4HhlH7L0kEh6mUq16db7++msgYyTu9OnTKV68OP369QNg/PjxLFy4kLNnz7Jt2zbKlSvHggULUBSFihUr8s8//zB69GjGjx+PRpPxPYurqyszZsxQzzllyhRq1qzJ1KlT1W1Lly6lXLlyXLp0CTc3t1d4B8SrlpiYyM8//8zKlSvVETLLly+nbNmy2baPjo7Gzs6OJk2aYGRkhIODA7Vr1wbg0qVL/Prrr+zZs4cmTZoA4OzsnKWPSZMm0bRpUyBj5NnUqVPZu3ev+tnB2dmZw4cPs2jRIho1aoStbUZNwpIlS6o1yXJzXF6IiYmhS5cuzJgxg3bt2nHv3j0OHTqkjlTat28fpqamBAUFERUVRa9evShWrBjffPMNkDFK74svvqBatWokJiYyfvx42rVrx+nTp9FoNCQmJtKoUSPKlCnD1q1bsbOzIzQ0VK3HdejQIbp3746/vz8NGjTgypUr9O+fMWJrwoQJatJy7dq1VK5cmRs3bnDmzJk8ufbsSB1JIYQQQrxOXmqSDOCLL76gR48eeHl5Ubt2bebOnatO7YGMb4jLlCnDtGnTABgyZAiNGjVi9uzZtGzZkrVr13Ly5EkWL178skMVQohCJfB8DH7bwtV6PzdiErAp7Uzg+Rh8qthjYGBAsWLF9EbtZNbziY2NJSIigrp166Io/9Vvq1+/PomJifz99984ODgA8NZbb+md98yZMxw4cCDLaB+AK1euSJKskMpMyB47GUpKSgpetWqr+2xtbXF3d8/2uI4dOzJ37lycnZ3x8fGhRYsWtG7dGkNDQ06fPo2BgcFTE1ReXl7q73/++ScPHjxQk2aZUlJS1Clv2Xne43Lj0WT1nWsXSUtLo3379mqx9kf/HzQ2Nmbp0qWYm5tTuXJlJk2axMiRI5k8eTIajYYOHTro9b106VJKlChBeHg4VapUYfXq1dy8eZMTJ06oyUAXFxe1vZ+fH2PGjKFHjx5ARiJw8uTJjBo1igkTJjwxafkyZNaRHLAyFAX0EmVSR1IIIYQQr9pLT5J16tSJmzdvMn78eG7cuEGNGjUIDAxUH8Sio6PV0QgA9erVY/Xq1Xz99dd89dVXuLq6snnzZqpUqfKyQxVCiEIj8HwMA1aGZhmZ8SANBqwMVWuzKYqCkZGRuj8zIfYsq8A9vkphYmIirVu35ttvv83S1t7+zZ7uWlg9mpBNib0KQIeFIUztZvzUKc7lypXj4sWL7N27lz179jBw4EBmzpxJcHAwZmZmuTr/o/8GExMTAdixYwdlypTRa2diYpJjH8973NM8nqzWadOxqlCTSpWr0LK5D++//z4ffvghRYsWBaB69ep6ixrUrVuXxMRE/vrrLxwdHbl8+TLjx4/n2LFj3Lp1S/1/NTo6mipVqnD69Glq1qypJsged+bMGUJCQtSRaQDp6ekkJSXx4MGDJyYtX5bMOpKP3ifIGEE2obXHGz9NXgghhBCvzktPkgH4+vrmOL0yKCgoy7aOHTvSsWPHlxyVEEIUTulaHX7bwrOdupTJb1s4TT2yr/WYqVKlSmzYsAGdTqcmz0JCQrC0tMxx6hyAp6cnGzZswMnJ6aU+WIvXw+MJWUMbe9AY8vfFcwxYac3CTzypU8aUS5cu5TgizMzMjNatW9O6dWs+//xzKlasyLlz56hatSparZbg4GB1uuXTeHh4YGJiQnR0dI7nMzY2BjKSQ89y3LPKLlmtaAwo2mESydcjMLaMYf78+YwdO1avZtiTtG7dGkdHR5YsWULp0qXRarVUqVJFrVv2tMRiYmIifn5+tG/fPss+U1PTJyYtH02o5zWfKvY09bBTR9yVtMyYYikjyIQQQgjxKsnTixBCFDLHI+P0RmM8TgfExCdxPDLuif0MHDiQuXPnMmjQIHx9fbl48SITJkzgiy++0BsB/LjPP/+cJUuW0KVLF0aNGoWtrS1//vkna9eu5aeffsLAwOB5L028ZrJLyGqMzbCo1pS4A0vRmFkyekkMjpHbcvw3ExAQQHp6OnXq1MHc3JyVK1diZmaGo6MjxYoVo0ePHvTu3Rt/f3+qV6/OtWvXiI2NzXEhCEtLS0aMGMGwYcPQarW88847xMfHExISgpWVFT169MDR0RFFUdi+fTstWrTAzMwsV8e96L1RKQqmZT24au3JSf8ZOJd3YtOmTUDGSK+HDx+qya6jR49iYWFBuXLluH37NhcvXmTJkiU0aNAAgMOHD+t1Xa1aNX766Sfi4uKyHU3m6enJxYsX9aZgPi6npKWnp+cz3YNnJXUkhRBCCJHfJEkmhBCFTOy9nBNkz9KuTJky7Ny5k5EjR1K9enVsbW3p06ePWvg/J6VLlyYkJITRo0fz/vvvk5ycjKOjIz4+Pk9MromCJ6eEbNF3e6NLTSJ2wyRuGpvRwHcIbyXEZ9uHjY0N06dP54svviA9PZ2qVauybds2ihXLSJYsXLiQr776ioEDB3L79m0cHBz46quvnhjX5MmTKVGiBNOmTePq1avY2Njg6empHlemTBm1NlevXr3o3r07AQEBTz0uL+5N8j8XSbp2BlOnmvyVYM3MH3/h5s2bVKpUibNnz5KSkqL+fxYVFcWECRPw9fVFo9FQtGhRihUrxuLFi7G3tyc6OpoxY8bo9d+lSxemTp1K27ZtmTZtGvb29oSFhVG6dGnq1q3L+PHjadWqFQ4ODnz44YdoNBrOnDnD+fPnmTJlyhOTlkIIIYQQhZ2iy1xOqZBISEjA2tqa+Ph4rKys8jscIYR45Y5cuU2XJUef2m5Nv7dl1IZ4IVtOX2fI2tNPbTevcw3a1Cjz1HaFSU73JvXWX8TtX0LKv1fQJj+gdNlyfDliGL6+vvTs2ZO7d+9SvXp1vv/+e5KTk+nSpQvz589X66Lt3buXwYMHc/XqVdzd3fH398fb25tNmzapK4lfu3aN4cOHs2fPHtLS0vDw8OD7779XC/Dv2rWLSZMmERYWhpGRERUrVqRv377069ePzZs3M336dCIiItSk5ZQpU9SVSoUQQgghCqLc5ookSSaEEIVMulbHO9/u50Z8UrZTvRQyCmIfHv2e1PsRL0QSsjl7nnuTmSTbvHnzS45OCCGEEOLNkttckcx7EUKIQsZAozChtQeQkRB7VObrCa09JEEmXljt8rbYW5tm+XeWSQHsrTMKsL9p5N4IIYQQQhQ8kiQTQohCyKeKPQs/8cTO2lRvu521KQs/8cSnin0+RSYKE0nI5kzujRBCCCFEwSPTLYUQohBL1+o4HhlH7L0kSlpmjFqRh3KR1wLPx+C3LVyvUL29tSkTWnu88QlZuTdCCCGEEPlPapJJkkwIIYR4ZSQhmzO5N0IIIYQQ+Su3uSLDVxiTEEIIIQopA43yxhXnzy25N0IIIYQQBYPUJBNCCCGEKKQURcHU1BRTU1MURUFRFKytrUlLS8vv0IQQQgghXjuSJBNCCCGEKMSSk5NRFIVJkyZRuXJlEhIScHNzy++whBBCCCFeOzLdUgghhBCiEElMSmPYujCi7zxUt927dw9DQ0PGjRuHhYUFkZGR+RihEEIIIcTrSUaSCSGEEEIUEh8sOESVibvYExHLxRv3MjZqDGj/4xG1TbVq1QBITEzMjxCFEEIIIV5bkiQTQgghhCgEPlhwiLN/J2S77+zfCXyw4NArjkgIIYQQomCRJJkQQgghRAGXmJSWY4IMbTqQkShLTErj3LlzAFhYWLyq8IQQQgghCgRJkgkhhBBCFHDD1oU9cf+12e25G7IGB9eKJCYm4ujo+IoiE0IIIYQoOKRwvxBCCCFEAfdokf4sDIxAqyX+8CogYwTZn3/++YoiE0IIIYQoOCRJJoQQQghRwDkUNfuvUH8WCo4jNwHQtFJJlvSo9eoCE0IIIYQoQGS6pRBCCCFEATenU808bSeEEEII8SaSJJkQQgghRAFnYWpItbJWT2xTrawVFqYyiUAIIYQQIieKTqfT5XcQeSkhIQFra2vi4+Oxsnryh0UhhBBCiMLkgwWHsl3lslpZK7b6NsiHiIQQQggh8l9uc0XydaIQQgghRCGx1bcBiUlpDFsXRvSdhzgUNWNOp5oygkwIIYQQIhfkE5MQQgghRCFiYWooxfmFEEIIIZ6D1CQTQgghhBBCCCGEEG88SZIJIYQQQgghhBBCiDeeJMmEEEIIIYQQQgghxBtPkmRCCCGEEEIIIYQQ4o0nSTIhhBBvHEVR2Lx5c36HIYQQQgghhHiNSJJMCCFEgSCJLSGEEEIIIcTLJEkyIYQQQgghhBBCCPHGkySZEEK8AbRaLTNmzMDFxQUTExMcHBz45ptvADh37hzvvfceZmZmFCtWjP79+5OYmKge27NnT9q2bcusWbOwt7enWLFifP7556SmpqptnJycmDp1Kr1798bS0hIHBwcWL16sF8Nff/3FRx99hI2NDba2trRp04aoqCi9NkuXLqVy5cqYmJhgb2+Pr6+v2j9Au3btUBRFfQ2wZcsWPD09MTU1xdnZGT8/P9LS0tT9ly9fpmHDhpiamuLh4cGePXv0zhkVFYWiKGzcuJF3330Xc3NzqlevzpEjR/TaHT58mAYNGmBmZka5cuUYPHgw9+/fV/f/8MMPuLq6YmpqSqlSpfjwww/Vfb/99htVq1ZV73GTJk30jhVCCCGEEELkP0mSCSHEG+DLL79k+vTpjBs3jvDwcFavXk2pUqW4f/8+zZo1o2jRopw4cYL169ezd+9eNTmV6cCBA1y5coUDBw6wfPlyAgICCAgI0Gsze/ZsvLy8CAsLY+DAgQwYMICLFy8CkJqaSrNmzbC0tOTQoUOEhIRgYWGBj48PKSkpACxcuJDPP/+c/v37c+7cObZu3YqLiwsAJ06cAGDZsmXExMSorw8dOkT37t0ZMmQI4eHhLFq0iICAADUBqNVqad++PcbGxhw7dowff/yR0aNHZ3uPxo4dy4gRIzh9+jRubm506dJFTbZduXIFHx8fOnTowNmzZ1m3bh2HDx9W79PJkycZPHgwkyZN4uLFiwQGBtKwYUMAYmJi6NKlC7179yYiIoKgoCDat2+PTqd7of+mQogXFxQUhKIo3L17N79DEUIIIcTrQFfIxMfH6wBdfHx8focihHiFjh07pgP0ftzc3HQ6nU5XqlQpHaCzsrLS21+7dm29PsqWLZulj82bN6v7AZ2iKHr7DQwMXul1Pou0dK3ujz9v6dYc/p/O2NhEt2jR4ixtFi9erCtatKguMTFR3bZjxw6dRqPR3bhxQ6fT6XQ9evTQOTo66tLS0tQ2HTt21HXq1El97ejoqPvkk0/U11qtVleyZEndwoULdTqdTrdixQqdu7u7TqvVqm2Sk5N1ZmZmul27dul0Op2udOnSurFjx+Z4PYBu06ZNetsaN26smzp1qt62FStW6Ozt7XU6nU63a9cunaGhoe769evq/t9//12vr8jISB2g++mnn9Q2Fy5c0AG6iIgInU6n0/Xp00fXv39/vfMcOnRIp9FodA8fPtRt2LBBZ2VlpUtISMgS96lTp3SALioqKsdrE0K8Go0aNdINGTJEfX3gwAEdoLtz506+xSSEEEKIly+3uSIZSSaEKBTq1KkDgLW1NfXq1QPg0qVLzJo1S22TkJCAra0tXl5eABw/fpz//e9/ALRq1Yq///4bRVF4//33cXBwAKBt27Z659HpdGg0Gpo3b46iKKSnp9O4ceOXfXnPLPB8DO98u58uS44ybNEOUlKSWXjZjMDzMXrtIiIiqF69OkWKFFG31a9fH61Wq44CA6hcuTIGBgbqa3t7e2JjY/X6qlatmvq7oijY2dmpbc6cOcOff/6JpaUlFhYWWFhYYGtrS1JSEleuXCE2NpZ//vnnme/lmTNnmDRpktqnhYUF/fr1IyYmhgcPHhAREUG5cuUoXbq0ekzdunWz7evR+O3t7QH04g8ICNA7T7NmzdBqtURGRtK0aVMcHR1xdnamW7durFq1igcPHgBQvXp1GjduTNWqVenYsSNLlizhzp07z3SdQgghhBBCiJdPkmRCiALpxt0kvCbvxm3sTpy7TATAxMSEu3fvEhISQkREBACjRo1SjzExMeH27ducOHGCmTNnAlCzZk0AduzYAWRMz9u1axfXrl2jbNmyQEYtqkelp6ezc+dONQkSFBT00q7zeQSej2HAylBi4pMAUIxMALh5L5kBK0OzJMpyw8jISO+1oihotdpct0lMTOStt97i9OnTej+XLl3i448/xszM7JljyuzXz89Pr89z585x+fJlTE1Nn6mvR+NXFAVAL/5PP/1U7zxnzpzh8uXLVKhQAUtLS0JDQ1mzZg329vaMHz+e6tWrc/fuXQwMDNizZw+///47Hh4ezJ8/H3d3dyIjI5/rmoUQz6dnz54EBwczb948FEVBURS1LuKpU6fw8vLC3NycevXq6X1JAE+ufajT6Zg4cSIODg6YmJhQunRpBg8erB6bnJzMiBEjKFOmDEWKFKFOnTqv3d8NIYQQQmSQJJkQosCpNO533p6+j1v3U0lJ1/HX/pUAGFWoo7apWLEigF7dp44dO6q/jxgxAsh4eHlU5oOToij8/fffAHz11Vd6+zNlJmF0r1FtqXStDr9t4TwakVHR0iiGJjy8dgYAv23hpGszWlSqVIkzZ87oFZEPCQlBo9Hg7u6eZ3F5enpy+fJlSpYsiYuLi96PtbU1lpaWODk5sW/fvhz7MDIyIj09PUu/Fy9ezNKni4sLGo2GSpUq8ddffxET819i8OjRo88Vf3h4eLbnMTY2BsDQ0JAmTZowY8YMzp49S1RUFPv37wcy/t3Ur18fPz8/wsLCMDY2ZtOmTc8chxDi+c2bN4+6deuqo01jYmIoV64ckFGTcPbs2Zw8eRJDQ0N69+6tHve02ocbNmxgzpw5LFq0iMuXL7N582aqVq2qHu/r68uRI0dYu3YtZ8+epWPHjvj4+HD58uVXewOEEEII8VSG+R2AEEI8i0rjfudhqjbbfWnpGfsjJjd/7v6bN8967NSpU594zOuUJDseGaeOIMukGBpjVacDd4OWoRgYklLGg6Ubd6OJ/5uuXbsyYcIEevTowcSJE7l58yaDBg2iW7dulCpVKs/i6tq1KzNnzqRNmzZMmjSJsmXLcu3aNTZu3MioUaMoW7YsEydO5LPPPqNkyZI0b96ce/fuERISwqBBgwDUJFr9+vUxMTGhaNGijB8/nlatWuHg4MCHH36IRqPhzJkznD9/nilTptCkSRPc3Nzo0aMHM2fOJCEhgbFjxz5z/KNHj+btt9/G19eXvn37UqRIEcLDw9mzZw8LFixg+/btXL16lYYNG1K0aFF27tyJVqvF3d2dY8eOsW/fPt5//31KlizJsWPHuHnzJpUqVcqz+yuEyF66VsfxyDhi7yVR0tIUI2NjzM3NsbOzA1Cn3H/zzTc0atQIgDFjxtCyZUuSkpIwNTXFz8+PMWPG0KNHDwCcnZ2ZPHkyo0aNYsKECURHR2NnZ0eTJk0wMjLCwcGB2rVrAxAdHc2yZcuIjo5Wp32PGDGCwMBAli1b9tS/L0IIIYR4tSRJJoQoMG7cTco2QWZdrzO3N39D0uU/eJiq5cbdJO7eiAL0R3799ttvrFixAgB/f38gYwrmo3bu3PmSon81Yu8lZbvdun5nFI0Bdw+tIj0xjrFbSzFs8OeYm5uza9cuhgwZQq1atTA3N6dDhw589913eRqXubk5Bw8eZPTo0bRv35579+5RpkwZGjdujJWVFQA9evQgKSmJOXPmMGLECIoXL86HH36o9jF79my++OILlixZQpkyZYiKiqJZs2Zs376dSZMm8e2332JkZETFihXp27cvABqNhk2bNtGnTx9q166Nk5MT/v7++Pj4PFP81apVIzg4mLFjx9KgQQN0Oh0VKlSgU6dOANjY2LBx40YmTpxIUlISrq6urFmzhsqVKxMREcHBgweZO3cuCQkJODo6Mnv27GwTskKIvBN4Pga/beF6XxzERd+haLn7WdrmVJPQwcGBM2fOEBISoo4cg4xp90lJSTx48ICOHTsyd+5cnJ2d8fHxoUWLFrRu3RpDQ0POnTtHeno6bm5ueudLTk6mWLFieX3JQgghhHhBiu51GgKRBxISErC2tiY+Pl598BJCFA5ek3dz635qtvuufdsKAMWkCGYlHXjwV0ZNsnnz5jF16lT+/fdfAIoXL46zszPHjx8HICwsjBo1avDWW28RGhoKZBRaL1WqFCEhIdy/f18dKZY5DfPRWlyZSbjX5a30yJXbdFny9OmEa/q9Td0K8oAmhCicMmszPv7OfGP1GIxLOvNrwI/4VLEnKCiId999lzt37mBjYwNkfFnSsmVLduzYQYsWLTA2NiY1NZVTp05l+Wzp7OyMRqPh4cOH7N27lz179rB+/XrKly9PcHAwGzdupGvXrly4cEFv8RMACwsLdUSbEEIIIV6u3OaKpCaZEKLASEhKy3FfqS7TAdAl31cTZG5ubnrFk62srLh165aaIPP09KRGjRpARtHmzG/6z5w5w+7du7l//77eSLSCoHZ5W+ytTckpagWwtzaldnnb5+pfURQ2b978vOFl4e3tzdChQ/OsPyGEyK42YybFwAh0Wr3ajE+T+bfB2dk529qHAGZmZrRu3Rp/f3+CgoI4cuQI586do2bNmqSnpxMbG5vlWEmQCSGEEK8fSZIJIQoMK9OcZ4ibOlTBcfR2HEdv561Ju9DpdFlWJ5s2bRo6nU79OXXqlN7+ixcv6u3X6XR6o8Yef5257XUZRQZgoFGY0NoDIEuiLPP1hNYeGGienPybOHGimkB8VExMjEwTFEK81rKrzZjJ0LokyTEX+Sv6GrtPXc7ynp6d7t27AzB9+nQuXLhAREQEa9eu5euvvwYgICCAn3/+mfPnz3P16lVWrlyJmZkZjo6OuLm50bVrV7p3787GjRuJjIzk+PHjTJs2TV1VWQghhBCvD6lJJoQoMLYPasjb03Ne/fDRdm8ynyr2LPzEM0stHjtrUya09sCnin2Ox+p0uiwrSD7qdRv5kBmvoaH8ORNCZMipNiOAVe323NrxHdcXf0aLH1MxMzMDoFOnTvzwww9UqFAhyzGZRfgPHDiAv78/RkZGlCxZkvv37zNz5kysra0xMzPj7t27pKenU6JECUqVKqXWHGvbti2rVq2iX79+3Lt3j+LFi5OWlkbbtm1p2bIlZ86cYejQoZw8eRJFUXB1dWXRokV4eXm9hLsjhBBCiCeRkWRCiALDzsYUM6Mnv22ZGWmwszF9RRHlP29vb3x9ffH19cXa2prixYszbtw4mlW24/Do9+hePBrDrV9xc0EnIud15Zepw4mNjVWPDwoKQlEUfv/9d9566y1MTExYuXIlfn5+nDlzRq3DFhAQAGSdbvn333/TpUsXbG1tKVKkCF5eXhw7dgyAnj170rZtW714hw4dire3d47Xs2LFCry8vLC0tMTOzo6PP/74qfEePnz4he+jEKLwKGmZ898AI9sy2HebTfEPRjD1+wDOnj1LWFgYJiYmtGvXDq1Wi4dHxmjczNUoM+3atYsHDx6wf/9+rl69yoABAzh37hwzZszg5s2bzJs3j8TERLZu3cq1a9e4efMmACEhIRQvXpymTZuSkpLCtWvXuH//Ph999BGQsfpv2bJlOXHiBKdOnWLMmDEYGRm9pLsjhBBCiCeRr96FEAVKxOTmVBr3e7arXJoZaYiYnHUq4I0bN15FaPlm+fLl9OnTh+PHj3Py5En69++Pg4MD/fr1w8nWhPmzp+Pu7k5sbCxffPEFPXv2zLKK55gxY5g1axbOzs6YmpoyfPhwAgMD2bt3LwDW1tZZzpuYmEijRo0oU6YMW7duxc7OjtDQ0FxNX8pJamoqkydPfqZ4ixYt+tznE0IUPpm1GW/EJ2Vflwxwqd2YUZ+9p049X7p0KSVKlCA8PBwLC4sn9v/dd9/RuHFjxo0bB2TULAsPD2fmzJn07NmTKlWqYGtrS3BwMB9++CFBQUEMHz6cefPmAXD8+HFSU1OpV68eANHR0YwcOZKKFSsC4Orqmjc3QgghhBDPTJJkQogCJ2Jyc27cTaLV/IMkJKVhZWrI9kEN35gRZOlaHccj44i9l0TCw1TKlSvHnDlzUBQFd3d3zp07x5w5c+jXrx+9e/dWj3N2dsbf359atWqRmJio9yA4adIkmjZtqr62sLDA0NDwidMrV69ezc2bNzlx4gS2thkLAbi4uLzQtT1vvEIIkSmzNuOAlaEokCVRpgN6VzHhk64fc+zYMW7duqUm96Ojo9WRZDmJiIigTZs2etvq16/P3LlzSU9Px8DAgIYNGxIUFESTJk0IDw9n4MCBzJgxg//9738EBwdTq1YtzM3NAfjiiy/o27cvK1asoEmTJnTs2DHbaZ9CCCGEePlkuqUQokCyszHl5Lj3ufRNC06Oe/+NSZAFno/hnW/302XJUYasPU14TAK3zB3YdeG/0XJ169bl8uXLpKenc+rUKVq3bo2DgwOWlpY0atQIyHgQfNTz1L45ffo0NWvWVBNkeeFlxiuEeHNk1ma0s87+b8Pg3h9zKTqGJUuWcOzYMXWaeEpKSp6c39vbm6CgIA4dOkTNmjWxsrJSE2fBwcHqextkLJRy4cIFWrZsyf79+/Hw8GDTpk15EocQQgghno0kyYQQooAIPB/DgJWhWVZte5iSzoCVoQSej9HbnpSURLNmzbCysmLVqlWcOHFCffB6/EGwSJEizxxPZsHrnGg0miwrf6ampubY/v79+y81XiHEm8Wnij3jWmYdFZb+MIGkW3/xj1NzUkt5UKlSJe7cuZPrfitVqkRISIjetpCQENzc3DAwMACgUaNGhIeHs379erUOo7e3N3v37iUkJCRLbUY3NzeGDRvG7t27ad++PcuWLXu2ixVCCCFEnpAkmRBCFADpWh1+28Kzra+T/M8lAPy2hZOu1XH06FFcXV353//+x+3bt5k+fToNGjSgYsWKekXwn8TY2PiJq1wCVKtWjdOnTxMXF5ft/hIlShATo5+4O336dI79vUi8QgjxuHStjsk7wrNs15haoDGz4t6ZXXwZsIc9e/fxxRdf5Lrf4cOHs2/fPiZPnsylS5dYvnw5CxYsYMSIEWqbatWqUbRoUVavXq2XJNu8eTPJycnUr18fgIcPH+Lr60tQUBDXrl0jJCSEEydOUKlSpRe7eCGEEEI8F0mSCSFEAXA8Mi7LCLJMafducnvfEqKv/skU/yXMnz+fIUOG4ODggLGxMfPnz+fq1ats3bqVyZMn5+p8Tk5OREZGcvr0aW7dukVycnKWNl26dMHOzo62bdsSEhLC1atX2bBhA0eOHAHgvffe4+TJk/zyyy9cvnyZCRMmcP78+RzP+SLxCpEb3t7eDB06NNftN2/ejIuLCwYGBgwdOpSAgABsbGxe6jlF3snpfVNRNBT/YBQpN/7k9Ny+DBw0hJkzZ+a6X09PT3799VfWrl1LlSpVGD9+PJMmTaJnz56PnEOhQYMGKIrCO++8A2QkzqysrPDy8lJHwxoYGHD79m26d++Om5sbH330Ec2bN8fPz+/FLl4IIYQQz0UK9wshRAEQey/7BBlAkcrvoUtLIeaXL5hlasSQIUPo378/iqIQEBDAV199hb+/P56ensyaNYsPPvjgqefr0KEDGzdu5N133+Xu3bssW7ZM7wEQMkab7d69m+HDh9OiRQvS0tLw8PDg+++/B6BZs2aMGzeOUaNGkZSURO/evenevTvnzp3L9pwlSpR47niFeBk+/fRTevXqxeDBg7G0tMTQ0JAWLVo8Ux8bN27EyMjoJUUonuRJ75tmTjUw67sQgFmda9CoRhm96eGP/u7t7Z1l6niHDh3o0KHDE8+/efNmvdcajSbLyFtjY2PWrFnzxH6EEEII8eoousf/6hdwCQkJWFtbEx8fj5WVVX6HI4QQeeLIldt0WXI0y/Ybq8dgXNIZ2yb9AVjT723qVij2qsMTokDw9vamRo0azJ0796ltExMTsbS0ZP/+/bz77rsvPziR53J633ycvG8KIYQQhV9uc0Uy3VIIIQqA2uVtsbc2RclhvwLYW5tSu3zerTQpREF2//59unfvjoWFBfb29syePVtvf3JyMiNGjKBMmTIUKVKEOnXqEBQUBEBQUBCWlpZAxrRhRVEICgrKMt1y4sSJ1KhRgxUrVuDk5IS1tTWdO3fm3r17apvHp1s6OTkxdepUevfujaWlJQ4ODixevFgvtj/++IMaNWpgamqKl5cXmzdvRlGUJ9b0E1nJ+6YQQgghnpUkyYQQogAw0ChMaJ2xSltOD3wTWntgoMlprxBvlpEjRxIcHMyWLVvYvXs3QUFBhIaGqvt9fX05cuQIa9eu5ezZs3Ts2BEfHx8uX75MvXr1uHjxIgAbNmwgJiaGevXqZXueK1eusHnzZrZv38727dsJDg5m+vTpT4xt9uzZeHl5ERYWxsCBAxkwYIB6voSEBFq3bk3VqlUJDQ1l8uTJjB49Oo/uypvlSe+bma/lfVMIIYQQj5IkmRBCFBA+VexZ+Ikndtam6ja7j6dTucNgFn7iiU8V+3yMToj8la7VceTKbbacvs6+s9f4+eefmTVrFo0bN6Zq1aosX76ctLQ0AKKjo1m2bBnr16+nQYMGVKhQgREjRvDOO++wbNkyjI2NKVmyJAC2trbY2dlhbGyc7Xm1Wi0BAQFUqVKFBg0a0K1bN/bt2/fEWFu0aMHAgQNxcXFh9OjRFC9enAMHDgCwevVqFEVhyZIleHh40Lx5c0aOHJmHd+rNkt37JoCdtam8bwohhBAiCyncL4QQBYhPFXuaethxPDKO2HtJlLTMmCokIyHEmyzwfAx+28LVlQxTYq+SkpLCQxsntY2trS3u7u4AnDt3jvT0dNzc3PT6SU5OplixZ6tN5eTkpE7NBLC3tyc2NvaJx1SrVk39XVEU7Ozs1GMuXrxItWrVMDX9L6lTu3btZ4pJ6JP3TSGEEELkliTJhBCigDHQKFJkWoj/F3g+hgErQ8luFaKxm85T0r5sltFCiYmJGBgYcOrUKQwMDPT2WVhYPNP5H1+5UlEUtFptnh8jXoy8bwohhBAiN2S6pRBCCCEKpHStDr9t4VkSZIY29qAxJPmfS/htCyddq+POnTtcunQJgJo1a5Kenk5sbCwuLi56P3Z2dq/+Qh7h7u7OuXPnSE5OVredOHEiHyMSQgghhHhzSJJMCCGEEAXS8cg4dYrlozTGZlhUa0rcgaVcPXuMNbtC6NmzJxpNxsceNzc3unbtSvfu3dm4cSORkZEcP36cadOmsWPHjld9GXo+/vhjtFot/fv3JyIigl27djFr1iwgY8SZEEIIIYR4eSRJJoQQQogCKfZe1gRZpqLv9sa0XGVubpjEoG7teeedd3jrrbfU/cuWLaN79+4MHz4cd3d32rZty4kTJ3BwcHgVoefIysqKbdu2cfr0aWrUqMHYsWMZP348gF6dMiGEEEIIkfcUnU6XXRmPAishIQFra2vi4+OxsrLK73CEEEII8ZIcuXKbLkuOPrXdmn5vF+h6VKtWraJXr17Ex8djZmaW3+EIIYQQQhQ4uc0VSeF+IYQQQhRItcvbYm9tyo34pGwL9yuAnXXGSoYFyS+//IKzszNlypThzJkzjB49mo8++kgSZEIIIYQQL5lMtxRCCCFEgWSgUZjQ2gPISIg9KvP1hNYeGGgKVi2vGzdu8Mknn1CpUiWGDRtGx44dWbx4cX6HJYQQQghR6Ml0SyGEEEIUaIHnY/DbFq5XxN/e2pQJrT3wqWKfj5EJIYQQQojXgUy3FEIIIcQbwaeKPU097DgeGUfsvSRKWmZMsSxoI8iEEEIIIUT+kiSZEEIIIQo8A41SoIvzC5EpKCiId999lzt37mBjY5Pf4QghhBBvFKlJJoQQQgghhBBCCCHeeJIkE0IIIYQQQgghhBBvPEmSCSGEEEKIQker1TJt2jTKly+PmZkZ1atX57fffkOn09GkSROaNWtG5vpVcXFxlC1blvHjxwOQnp5Onz591GPd3d2ZN2+eXv89e/akbdu2TJ06lVKlSmFjY8OkSZNIS0tj5MiR2NraUrZsWZYtW6YeExUVhaIorF27lnr16mFqakqVKlUIDg5+4rUcPnyYBg0aYGZmRrly5Rg8eDD379/P4zsmhBBCCEmSCSGEEEKIQmfatGn88ssv/Pjjj1y4cIFhw4bxySefcPDgQZYvX86JEyfw9/cH4LPPPqNMmTJqkkyr1VK2bFnWr19PeHg448eP56uvvuLXX3/VO8f+/fv5559/OHjwIN999x0TJkygVatWFC1alGPHjvHZZ5/x6aef8vfff+sdN3LkSIYPH05YWBh169aldevW3L59O9vruHLlCj4+PnTo0IGzZ8+ybt06Dh8+jK+v70u4a0IIIcSbTdFlfoVWSOR2WU8hhBBCCFF4pGt16gqnNsYKLWq5snfvXurWrau26du3Lw8ePGD16tWsX7+e7t27M3ToUObPn09YWBiurq459u/r68uNGzf47bffgIyRZEFBQVy9ehWNJuN754oVK1KyZEkOHjyYEVN6OtbW1vz000907tyZqKgoypcvz/Tp0xk9ejQAaWlplC9fnkGDBjFq1Kgshfv79u2LgYEBixYtUmM5fPgwjRo14v79+5iamub5vRRCCCEKm9zmimR1SyGEKAB0Oh2ffvopv/32G3fu3CEsLIwaNWrkd1hCCPFaCDwfg9+2cGLikwBIuXmNBw8e8F7jJhhoFLVdSkoKNWvWBKBjx45s2rSJ6dOns3DhwiwJsu+//56lS5cSHR3Nw4cPSUlJyfK+W7lyZTVBBlCqVCmqVKmivjYwMKBYsWLExsbqHfdo4s7Q0BAvLy8iIiKyvbYzZ85w9uxZVq1apW7T6XRotVoiIyOpVKlSbm6REEIIIXJBkmRCCFEABAYGEhAQQFBQEM7OzhQvXjy/QxJCiNdC4PkYBqwM5dGpEbrUjGSZTdtxfPNJQxq6lVT3mZiYAPDgwQNOnTqFgYEBly9f1utz7dq1jBgxgtmzZ1O3bl0sLS2ZOXMmx44d02tnZGSk91pRlGy3abXa576+xMREPv30UwYPHpxln4ODw3P3K4QQQoisJEkmhBAFwJUrV7C3t6devXrPdbxOpyM9PR1DQ3nbF0IUHulaHX7bwnm8dohRsXJgYERawk2WnHlID58KeiPKAIYPH45Go+H333+nRYsWtGzZkvfeew+AkJAQ6tWrx8CBA9X2V65cybO4jx49SsOGDYGM6ZanTp3KscaYp6cn4eHhuLi45Nn5hRBCCJE9KdwvhBCvuZ49ezJo0CCio6NRFAUnJyeSk5MZPHgwJUuWxNTUlHfeeYcTJ06oxwQFBaEoCr///jtvvfUWJiYmHD58OB+vQgghnp23tzdDhw7Ncf/xyDh1iuWjNCbmWNVuT9z+n7h8eAcbg04RGhrK/PnzWb58OTt27GDp0qWsWrWKpk2bMnLkSHr06MGdO3cAcHV15eTJk+zatYtLly4xbtw4vffYF/X999+zadMm/ve///H5559z584devfunW3b0aNH88cff+Dr68vp06e5fPkyW7ZskcL9QgghxEsgSTIhhHjNzZs3j0mTJlG2bFliYmI4ceIEo0aNYsOGDSxfvpzQ0FBcXFxo1qwZcXFxeseOGTOG6dOnExERQbVq1fLpCoQQ4uWIvZc1QZbJpsEnWNfrRPzR9XzcrB4+Pj7s2LEDJycn+vTpw8SJE/H09ATAz8+PUqVK8dlnnwHw6aef0r59ezp16kSdOnW4ffu23qiyFzV9+nSmT59O9erVOXz4MFu3bs1xGn21atUIDg7m0qVLNGjQgJo1azJ+/HhKly6dZ/EIIYQQIoOsbimEEK+hR1dpK2lpypGtv+A/bx5RUVHcv3+fokWLEhAQwMcffwxAamoqTk5ODB06lJEjR6qro23evJk2bdrk89UIIcTz8fb2pkaNGsydOzfb/Ueu3KbLkqNP7WdNv7epW6FYHkf37DJXt5TFV4QQQohXS1a3FEKIAurxVdoAOH+Zh6npQEZdnNTUVOrXr6/uNjIyonbt2llWR/Py8nolMQshxMui1WoZNWoUP/30E8bGxnz22WdMnDgRgJDNAcQGLCApLgaNqSVmLrUp6t0LjbEZAOnxsSQGLabFov+RkpKCk5MTM2fOpEWLFvl4RUIIIYR4Xcl0SyGEeI1krtL2eI2dhIdp3E5MIfB8zDP1V6RIkbwMT4hc8/b2ZvDgwYwaNQpbW1vs7OzUxAZAdHQ0bdq0wcLCAisrKz766CP+/fff/AtYvLaWL19OkSJFOHbsGDNmzGDSpEns2bMHAEMDAyZMnUnpPj9QvOUwkq6d5U7QMgAU4PaehZQvaszBgwc5d+4c3377LRYWFvl4NUIIIYR4nUmSTAghXhM5rdL2KL9t4TiVd8bY2JiQkBB1e2pqKidOnMDDw+PlBypELuWU3NBqtbRp04a4uDiCg4PZs2cPV69epVOnTvkdsngNVatWjQkTJuDq6kr37t3x8vJi3759AAwdOpTRvT/kp8+bU75aHWwafMKD/2UsUmJnbUopTSJtmr1H1apVcXZ2plWrVuqqkvnByckJnU4nUy2FEEKI15RMtxRCiNdETqu0PSomPokLsckMGDCAkSNHYmtri4ODAzNmzODBgwf06dPnFUUrRFaP1tJLeJhK1f9PbkDGaoELFixQkxvnzp0jMjKScuXKAfDLL79QuXJlTpw4Qa1atfLtGkT+erweow6yLDpib29PbGwsAHv37mXatGn873//IyEhgZTUNLTJSSz7pBoNPcqyrMQXDBgwgN27d9OkSRM6dOggi5gIIYQQIkeSJBNCiNfEk1Zpe7zd9OnT0Wq1dOvWjXv37uHl5cWuXbsoWrToS45SiOw9XkvvRkwCNqWdCTwfg08Ve+C/5EZERATlypVTE2QAHh4e2NjYEBERIUmyN1R29Rjjou9QtFyKXjtFUdBqtURFRdGqVSsGDBjAN998g62tLYcPH6ZPnz7ULGuJgUahb9++NGvWjB07drB7926mTZvG7NmzGTRo0Ku+PCGEEEIUADLdUgghXhMlLU1z3GdVqw1lByxV25mamuLv78/NmzdJSkri8OHDeokFb29vdDodNjY2LztsIXKspfcgDQasDFVr6WUmN4R4XE7/hlLStOyPiM22HuOpU6fQarXMnj2bt99+Gzc3N/75558s7cqVK8dnn33Gxo0bGT58OEuWLHlp1yGEEEKIgk2SZEII8ZqoXd4We2tTlBz2K4C9tSm1y9u+yrCEeKLc1tJL1/7XolKlSvz111/89ddf6rbw8HDu3r0rdfXeQM/zbwjAxcWF1NRU5s+fz9WrV1mxYgU//vijXpuhQ4eya9cuIiMjCQ0N5cCBA1SqVOklXIUQQgghCgNJkgkhxGvCQKMwoXVGguDxRFnm6wmtPTDQ5JRGE+LVe1otPR0ZtfSOR8ap25o0aULVqlXp2rUroaGhHD9+nO7du9OoUSO8vLxeQdTidZLbeoyP/hsCqF69Ot999x3ffvstVapUYdWqVUybNk2vTXp6Op9//jmVKlXCx8cHNzc3fvjhhzy/BiGEEEIUDlKTTAghXiM+VexZ+Ilnlro8dtamTGjtodZ2EuJ18Sy19DIpisKWLVsYNGgQDRs2RKPR4OPjw/z5819WmOI19qR/Q3YfT8/SbvPmzeq2YcOGMWzYML1junXrpv4u/6aEEEII8SwkSSaEEK8Znyr2NPWw01vhrXZ5WxlBJl5LOdXSezS5kdnu0eSGg4MDW7ZseZmhiQLiSfUYn6edEEIIIcTzkiSZEEK8hgw0CnUrFMvvMIR4qsxaejfik7KtKaWQMRJSaumJnMi/ISGEEEK8LqQmmRBCCCGe27PW0jM3N0dRFPVHCKnHKIQQQojXhSTJhBBCCPFCMmvp2VnrT4ezszZl4SeeerX0Hj58CMA777yTpZaUeHM9y78hIYQQQoiXRdHpdE9acbvASUhIwNramvj4eKysrPI7HCGEEOKNka7VPbWWXubosUL28UPkkdz8GxJCCCGEeFa5zRVJTTIhhBBC5AkDjcLNCyG0bdNGb7uVlRXOzs6cPn1a3SbJMpEdqccohBBCiPwk0y2FEEIIkWfa/H+CTKPRUKZMGSDjmztPT09atmyptmvZsiXLly/PlxiFEEIIIYTIjowkE0IIIcQL+fNGIs39g7l5dIu6LT09HYCQkBDeeecdli5dik6nU0eQbd++PV9iFUIIkT8URcHU1JSHDx+iKAp2dnbExMTkd1hCCKFHkmRCCCGEeG7lx+wgc8Jk/NF1ABiWcKH8mB1ETm9J/fr18y84IYQQr6Vt27bh6uqaq7aSUBNCvEoy3VIIIYQQz+XRBNnjdP+//3WSmpqa3yEIIYQAWrVqhbu7e36HIYQQWUiSTAghhBDP7M8biVkSZNZvdwIg7eafQEaibN3Wvc/V/+LFiyldujRarVZve5s2bejduzcAW7ZswdPTE1NTU5ydnfHz8yMtLU1tqygKCxcu5IMPPqBIkSJMmTIFFxcXZs2apdfn6dOnURSFP//887liFUIIoe/ChQsYGRmhKAqKolCqVCm9/YqiYG9vD0BaWho2NjZqW0VRMDMzU9sB3LhxQ90HsGfPHgwNDfWOqVatWpZzmJmZYWFhobaxsbHRa3P48GFMTEz0+mnYsKG6v1OnTlniunDhQp7eKyHE60WSZEIIIYR4Zs39g7Nss6n936qW12Z8QPT87nRu0xSAQYMGPVP/HTt25Pbt2xw4cEDdFhcXR2BgIF27duXQoUN0796dIUOGEB4ezqJFiwgICOCbb77R62fixIm0a9eOc+fO0adPH3r37s2yZcv02ixbtoyGDRvi4uLyTDEKIYTI3ltvvUVaWhotWrRgxIgR3Lx5M8e2derUIT4+nnfffZdFixbRq1cvNYG2b98+AGxsbFi/fj3r168HICYmBltbWwYPHsycOXMoXbo0586dU79EyZSUlISRkRHTpk3D3d2d+Ph4fHx81H0NGjQgJSWFZs2aMWfOHBo0aIChYUZFotmzZ/Prr79iZ2fHtGnT+Pjjj0lKSqJGjRp5fbuEEK8RRVfI1l5PSEjA2tqa+Ph4rKys8jscIYQQolByymEqZWL4QW5vm6G3zcrKivj4eOC/UQE5ffxI1+o4HhlH7L0kZo/sh0s5O5YuXQpkjC7z8/Pjr7/+4v3336dx48Z8+eWX6rErV65k1KhR/PPPP+q5hg4dypw5c9Q2//zzDw4ODvzxxx/Url2b1NRUSpcuzaxZs+jRo8dz3g0hhBCJSWkMWxfG+fPnODqrDzVq1CQsLBSA/fv307hx42wL99vZ2fHvv/9y8+ZNihcvnqXf3NYk02g0mJmZcf/+ffU40P97oygKRYoUITExkWbNmrF7926mTJnC2LFjs/RnbGyMoigkJyer2zp16sSvv/5KZGQkTk5Oz3yPhBD5J7e5IincL4QQQohnZqSBVG3W7RYeDbHwaKjX7vLUlurrJ303F3g+Br9t4cTEJwFwv0h1/lgzn7a+4/nA04lVq1bRuXNnNBoNZ86cISQkRG/kWHp6OklJSTx48ABzc3MAvLy89M5RunRpWrZsydKlS6lduzbbtm0jOTmZjh07Ptd9EEIIAR8sOMTZvxMAiAvaCsC1UvX4YMEhtvo24L333svx2G+++Ya+fftSokQJjIyMqFmzJrt27coyNfJRiYmJlCtXjrt37+ptfzShBWBiYqL3WqPRkJKSAkBoaEYCL7sEGfxXxzIz2faopUuXMmnSpBzjE0IUXJIkE0IIIcQz+31wI5rMzTrlMrt2uRF4PoYBK0P16pyZu9Tm9u86+k5ewuzP23Ho0CF1VFhiYiJ+fn60b98+S1+mpqbq70WKFMmyv2/fvnTr1o05c+awbNkyOnXqpCbVhBBCPJtHE2SPO/t3gpooy0mfPn1o1KgRvXr14uTJkxw/fpyiRYvmOLIMwMnJibt37+Lh4UHTpk0pU6YMo0ePzvJFTHYJrsw2xsbGT702U1PTbJNh3bp1e+qxQoiCSZJkQgghhHhmLnYWKJDj6pYAyv+3e5p0rQ6/beFZ+lIMjTF3q8f98CC+nn0Td3d3PD09AfD09OTixYvPVUesRYsWFClShIULFxIYGMjBgwefuQ8hhBAZUywfT5CZV2nMvZNbSAwLxKpGc87+ncDvu/c/sR8XFxcOHToEwOTJkxk/fjzjx4/nhx9+AMiyiEtcXBxGRkZqEf2kpCRGjRqVbVIsJ3Xq1GHTpk1888032Y4mMzAwIDk5mZEjR+a6TyFEwSdJMiGEEEI8l8jpLSk/Zke2iTLl//fnxvHIOHWK5eOKVPYm9jc//rkVTa8e/31zP378eFq1aoWDgwMffvihOgXz/PnzTJky5YnnMzAwoGfPnnz55Ze4urpSt27dXMUphBBC37B1YVm2mZZyBo0hqbFX+Pc3P4yKl6PFjE059uHu7k56ejodOnTA3t5eXYG4Xbt2QMZosFu3bvHLL79QsmRJfHx81NpmXbt2pUKFCkyfPv2ZY9+4cSMajYavv/6aQ4cO0bJlSzZt2oRGo2Hv3r0MGTKE7777DlNTUz799FPs7e3ZunUroaGhJCVl/zdLCFHwSZJMCCGEEM8tcnpL/ryRSHP/YFK1GTXIfh/cKFcjyDLF3sv5YcPUsRoGZpakxf3NW++1Urc3a9aM7du3M2nSJL799luMjIyoWLEiffv2zdU5+/Tpw9SpU+nVq1eu4xRCCKEv+s7DbLeX6v4d/y4fStKVEyRdOYFRERtS79/Ntq2trS1Hjx5lxoz/Fn1p0aIFTZtmrI78ySefsGLFCnVxFZ1Ox/bt22natCmrV68GMhaI0el0WUacPc3evXvx8fFh165d7Nq1C4CGDTPqas6ePZv4+HiWLl2Kv7+/XrxCiMJLVrcUQoh8FhAQwNChQ7MUnxXiTXHkym26LDn61HZr+r1N3QrF8uSchw4donHjxvz111+UKlUqT/oUQog3Tb/lJ9gTEfvUdk0rlWRJj1qvICIhhMhebnNFmlcYkxBCvPGcnJyYO3eu3rZOnTpx6dKl/AlIiNdA7fK22FubklMlGQWwtzaldvkX//Y+OTmZv//+m4kTJ9KxY0dJkAkhxAuY06lmnrYTQoj8JkkyIYTIhczlwl8GMzMzSpYs+dL6F+J1Z6BRmNDaAyBLoizz9YTWHhhocl+QOSdr1qzB0dGRu3fv6k3tEUII8ewsTA2pVvbJs3eqlbXCwlSq/AghCgZJkgkh3kje3t74+vri6+uLtbU1xYsXZ9y4ceqy4E5OTkyePJnu3btjZWVF//79ARg9ejRubm6Ym5vj7OzMuHHjSE1N1et727Zt1KpVC1NTU4oXL64WnvX29ubatWsMGzYMRVHUFZgCAgKwsbHJVR+QMRJmxIgRlClThiJFilCnTh2CgoJe0p0S4tXwqWLPwk88sbM21dtuZ23Kwk888alinyfn6dmzJ+np6Zw6dYoyZcrkSZ9CCPEm2+rbIMdEWbWyVmz1bfCKIxJCiOcnKX0hxBtr+fLl9OnTh+PHj3Py5En69++Pg4MD/fr1A2DWrFmMHz+eCRMmqMdYWloSEBBA6dKlOXfuHP369cPS0pJRo0YBsGPHDtq1a8fYsWP55ZdfSElJYefOnUDGKkrVq1enf//+6jmy86Q+AHx9fQkPD2ft2rWULl2aTZs24ePjw7lz53B1dX0Zt0qIV8Knij1NPew4HhlH7L0kSlpmTLHMixFkQgghXp6tvg1ITEpj2Lowou88xKGoGXM61ZQRZEKIAkcK9wsh3kje3t7ExsZy4cIFdUTXmDFj2Lp1K+Hh4Tg5OVGzZk02bcp5yXLISKStXbuWkydPAlCvXj2cnZ1ZuXJltu2dnJwYOnQoQ4cOVbc9Xrj/SX1ER0fj7OxMdHQ0pUuXVrc3adKE2rVrM3Xq1Ge5DUII8UYwMjJCq9WSnp4OgKIomJqa8vBh9ivz5YV27dqxefNmDh06xDvvvPPSziOEEEKIp8ttrkhS+0KIN0K6Vqc3OkUHvP3222qCDKBu3brMnj1bfYjy8vLK0s+6devw9/fnypUrJCYmkpaWpvcme/r06SeOEsuNJ/Vx7tw50tPTcXNz09uenJxMsWJ5s+qfEEKIZ5Nd0m3hwoV07dqVt99+Ox8jE0IIIcSzkCSZEKLQCzwfg9+2cGLik9RtcdF3MCn64InHFSlSRO/1kSNH6Nq1K35+fjRr1gxra2vWrl3L7Nmz1TZmZmYvHO+T+khMTMTAwIBTp05hYGCgt8/CwuKFzy2EECJv2NnZ8eGHH+Z3GEIIIYR4BlK4XwhRqAWej2HAylC9BBlASpqWoENHCDwfo247evQorq6uWZJPmf744w8cHR0ZO3YsXl5euLq6cu3aNb021apVY9++fTnGY2xsrI5Uy8mT+qhZsybp6enExsbi4uKi92NnZ/fEfoUQoiBLSkqiRIkS6sIniqKoI2hHjhyJRqNRtxsbG3P27Nlc9/33339TpEgRvb4fXTAFoEuXLnr7NRoNBw8exMjISI0vc9/3339Pu3btUBSFw4cPq33UqlVLr49SpUrpnSPzmoyNjdU25cuXf95bJoQQQohnJEkyIUShla7V4bctnJwKL6bdu0mvzwYRHvE/1qxZw/z58xkyZEiO/bm6uhIdHc3atWu5cuUK/v7+WWqWTZgwgTVr1jBhwgQiIiI4d+4c3377rbrfycmJgwcPcv36dW7dupXteZ7Uh5ubG127dqV79+5s3LiRyMhIjh8/zrRp09ixY8ez3SAhhChAihcvzq1bt3Bzc2PWrFl89NFH2NjYcOHCBWbNmoWiKIwaNYpWrVqRmprKW2+9leu+nZ2defDgAR9++CGzZs2iRIkSbN68menTpwMwZMgQ1q5di4mJCUOGDOHLL7+kdOnSPHjwgIMHDwIZdc/Wr1/P+vXr6dGjR5ZzDBgwgJMnT2JjY8OUKVNwc3MjNjYWd3d3vXZxcXHY2dkxY8YMihUrRlRUFJMnT36BOyeEEEKI3JLC/UKIQuvIldt0WXI02303Vo/BqLgj6LTo/jyMsZEhAwYMYMqUKSiKkm2BfYBRo0axdOlSkpOTadmyJW+//TYTJ05Ui+5DxiqWkydPJjw8HCsrKxo2bMiGDRuAjNFqn376KRcvXiQ5ORmdTpelcP/T+khNTWXKlCn88ssvXL9+neLFi/P222/j5+dH1apV8/QeCiHE6yA0NJS33noLZ2dnrly5orfP2dmZyMhIzp8/T+XKlYGMxVmCg4PZsGED7du3f2Lh/sWLF/Ppp5+yatUqPv74Y7VfjUZDkSJFuHfvHoaGhqSnp5PTx+bsapI9Xrjf0NAQrVaLVqtV21hYWHD//n21X0VRMDAwIC0tDYC0tDSMjIxwcHDIMnJZCCGEELmX21yRJMmEEIXWltPXGbL2dLb7bqweg3FJZ2yb9Gde5xq0qVHm1QYnhBDiiRKT0hi2LozoOw+J3DiLy4e24e/vz6BBg/TamZubk5SUpJd8OnLkCPXq1aNDhw789ttvT0ySNWrUSB0N9rjMhJWiKFhZWREfH59tu9wkyRRFwcbGhjt37qht6tWrx5EjR7h37x4WFhYoikKJEiWIjY3V6/vx44QQQgjxbGR1SyHEG6+kpWmethNCCPFqfLDgEGf/TlBfxyVlfGRdFPQnj+XIXlhm4mvSpEmYmur/PciPemDZ1cUsZN9pC/HMqlatyvnz5/n111/p2LFjfocjhCjEJEkmhCi0ape3xd7alBvxSTnWJbO3NqV2edtXGpcQQoicPZ4gA7DybMm9E5v4375f+WDBh2z1baDus7OzIzIykgsXLqjTLb/88ksAvemTOenYsSNnzpzh7NmzrF+/Pts2hoaGJCQkZLsPMkZ7PY2BgUGWkWiZiwvI6sRC6MtcuOLREaJ51a+JiQlJSUlPbyyEeCNJ4X4hRKFloFGY0NoDgMcfX+w/nk6xJv2Z0NoDA83TH26EEOJl8vb2zlID8U2UmJSWJUEGYGhjh2JShPT4G/w+oQtTp8+ke/fuuLi48PvvvwMZKwOPHj2aNm3aEBwcjKGhIe3bt3/qOceOHYuBgQG//fYbtWvXZv78+fTu3ZtixYrRqFEjACZOnAiAqakpw4cPx8/PD0dHRwIDAzPiMzQkKSmJ77//nsDAwGwfwPv3749Op6No0aJ88803uLu7c//+fdzc3J73dgkhhBAij0mSTIhcKugPMBMnTqRGjRr5HcYr51PFnoWfeGJnrT+Fxs7alIWfeOJTxT6fIhNCvC5SUlLeyHO/joatC8txX2nf5WhMLUiLu87YL0exYsUK7ty5g7u7OyNGjECn0zFjxgy2bt2KkZERp06dyvV5b9y4gZWVFSdOnGDw4MEsW7aMu3fvUrt2bSAjkdauXTuSk5P57rvvmDhxIn/99Zc6AmzChAkA+Pr60rx5c37++ecs5/jhhx/w8vLi7t27fP3111y6dIkSJUpw8eLFZ7lFQhR6mSMzdTqdOqIsOjoagHHjxqnbFEWhe/fu6nGfffaZ3j5FUfRWj83sNzk5Wd0vhBCPk8L9QuRSXFwcRkZGWFpaPrVtVFQU5cuXJyws7LVJTE2cOJHNmzdz+vTp/A4lX6RrdRyPjCP2XhIlLTOmWMoIMiEKJ29vb6pUqQLAihUrMDIyYsCAAUyaNEldvbZPnz5cvnyZzZs30759ewICAtiwYQPjx4/nzz//xN7enkGDBjF8+HC138zjwsPD2bp1KzY2Nnz11Vd8/vnnapu7d+8yYsQItmzZQnJyMl5eXsyZM4fq1asD/70X+/r68s0333Dt2jW6d+/O8uXL9a7h6tWrNG3alM8++4wRI0ao20+fPk3NmjW5fPkyLi4uL/M25otmcw9y8ca9p7Zzt7Nk19CGryAiIcSrtnnzZtq1awfAzJkzAfjpp5/UhHLVqlVxd3fnt99+A/6r2deiRQuCg4N5//33sbW1JSAgAK1Wy+jRo5k+fTorVqyge/fuGBoaMm3aNAC991chROGW21yRjCQTIpdsbW1zlSDLa6mpqa/8nIWRgUahboVitKlRhroVikmCTIhCbvny5RgaGnL8+HHmzZvHd999x08//aTunzVrFtWrVycsLIxx48Zx6tQpPvroIzp37sy5c+eYOHEi48aNIyAgQK/fmTNnqseNGTOGIUOGsGfPHnV/x44diY2N5ffff+fUqVN4enrSuHFj4uLi1DZ//vknGzZsYOPGjZw+fZp58+ZRt25d+vXrR0xMDDExMTg4ONC7d2+WLVumd/5ly5bRsGHDQpkgA3Aoapan7YQQBUNcYgrvfxdEDb/d/HDVBsgY+TVixAhGjBiBkZERAO7u7mr9QEdHR+C/2n47d+7k/v37bNq0iZ9//lldzXbevHkAdOvWDcioD5jZrxBCPE6SZELk0qPTLZ2cnJg6dSq9e/fG0tISBwcHFi9erLbNXA2rZs2aKIqCt7e3uu+nn36iUqVKmJqaUrFiRX744Qd1X1RUFIqisG7dOho1aoSpqSkLFy7EzMxMrbmSadOmTVhaWvLgwQMARo8ejZubG+bm5jg7OzNu3DhJsAkh3ljlypVjzpw5uLu707VrVwYNGsScOXPU/e+99x7Dhw+nQoUKVKhQge+++47GjRszbtw43Nzc6NmzJ76+vuoohkz169dnzJgxuLm5MWjQID788EO138OHD3P8+HHWr1+Pl5cXrq6uzJo1CxsbG3XEA2RMsfzll1+oWbMm1apVw9raGmNjY8zNzbGzs8POzg4DAwN69uzJxYsXOX78OJDxpcnq1avp3bv3K7iD+WNOp5p52k4I8fqrNWUPnlP2cCn2PncfpnIp9j6Q/aquQ4YMUX/PnEq5evVqAPbv349Go9GbbgkyrV0I8WwkSSbEc5o9ezZeXl6EhYUxcOBABgwYoA4Dz3yg2bt3LzExMWzcuBGAVatWMX78eL755hsiIiKYOnUq48aNyzLNJnN0QkREBB07dqRVq1bqB4BMq1atom3btpibmwNgaWlJQEAA4eHhzJs3jyVLlug9EAohRGGWrtVx5Mpttpy+TsLDVOrUqaNXb6Zu3bpcvnxZHVng5eWld3xERAT169fX21a/fn29YzL7eVTdunWJiIgA4MyZMyQmJlKsWDEsLCzUn8jISK5cuaIe4+joSIkSJZ56TaVLl6Zly5YsXboUgG3btpGcnEzHjh1zc0sKJAtTQ6qVfXK5jGplrbAwlQXahSgMak3Zw83EnJNYtabs0XtdvHhx9XeNJuNRNjk5GYDGjRuj0+mws7Pjgw8+oF+/fkD2yTYhhMiJfMIQ4gkerWOV8DBV749sixYtGDhwIJAximvOnDkcOHAAd3d39eGnWLFi2NnZqcdMmDCB2bNnq6ttlS9fnvDwcBYtWkSPHj3UdkOHDtVbkatr165069aNBw8eYG5uTkJCAjt27GDTpk1qm6+//lr93cnJiREjRrB27VpGjRqVx3dFCCFeL4HnY/DbFk5MfMaKgjdiEvg7PYbA8zE5Ls5RpEiRPI8jMTERe3t7goKCsuyzsbF5rnP37duXbt26MWfOHJYtW0anTp3UL0cKq62+DfhgwaFsV7msVtaKrb4N8iEqIURei0tMeWKCDOBmYgpxT2nzKENDQ2JiYoCMkWVLlizJ0kaSZkKIJ5EkmRA5yO6hK+bk3zQ/n/GHt1q1ampbRVGws7MjNjY2x/7u37/PlStX6NOnj/rNFkBaWhrW1tZ6bR8f4dCiRQuMjIzYunUrnTt3ZsOGDVhZWdGkSRO1zbp16/D39+fKlSskJiaSlpYmi1cIIQq9wPMxDFgZyuOPPHejIhiwMlRdxfbo0aO4urpiYGCQbT+VKlUiJCREb1tISAhubm56xxw9elSvzdGjR6lUqRIAnp6e3LhxA0NDQ5ycnJ7pOoyNjfVGrGVq0aIFRYoUYeHChQQGBnLw4MFn6reg2urbgMSkNIatCyP6zkMcipoxp1NNGUEmRCHSefEfT21ze+9i6v8ZhJLN+2N20tLSaN68OYqiZClVkiklJYWJEydSpkwZvc/kQggBMt1SiGxlPnRlJsgy3U9OY8DKUB6mpqsFRDMpioJWq82xz8TERACWLFnC6dOn1Z/z589neeh6fJSBsbExH374oTrlcvXq1XTq1AlDw4yHhSNHjtC1a1datGjB9u3bCQsLY+zYsVKDQQhRqKVrdfhtC8+SIANIu3eTuH1LGLN0F6tWrWb+/Pl6tWweN3z4cPbt28fkyZO5dOkSy5cvZ8GCBVkKO4eEhDBjxgwuXbrE999/z/r169V+mzRpQt26dWnbti27d+8mKiqKP/74g7Fjx3Ly5Mks55w4caK6ArKTkxPHjh0jKiqKW7duqX9PMmuTffnll7i6uvLll1+q9THzyqNxAPTs2ZO2bdvm6Tmeh4WpIUt61GLX0IYs6VFLEmRCFDKx93L+nGpeuTEAiae28r8Ns7h+/fpT+8t8bwwMDOT333/HzCzrAh9vv/02AH5+fvTv3/85ohZCFHbyaUOIxzzpoSvT3QepaJ8wVNvY2Dijr0e+9SpVqhSlS5fm6tWrdO3a9Znj6tq1K02bNuXChQvs37+fKVOmqPv++OMPHB0dGTt2rLrt2rVrz3wOIYQoSI5HxmX5MiNTkcrvoU1L4ez3nzPAzJghQ4Y88YHI09OTX3/9lfHjxzN58mTs7e2ZNGkSPXv21Gs3fPhwTp48iZ+fH1ZWVnz33Xc0a9YMyPiyZOfOnYwdO5ZevXpx8+ZN7OzsaNiwIaVKlcpyzhEjRjBo0CD19x49euDh4cHDhw+JjIxUR6P16dOHqVOn0qtXL/r06ZPlSxohhCiISloac/dh9otMlWg1DFoNA8CtZBF2f+Gdpc3jI8XmzJnz1Hq8R44ceb5ghRBvDEmSCfGYJz10AejISKRF336QY5uSJUtiZmZGYGAgZcuWxdTUFGtra/z8/Bg8eDDW1tb4+PiQnJzMyZMnuXPnDl988cUT42rYsCF2dnZ07dqV8uXLU6dOHXWfq6sr0dHRrF27llq1amWpVyaEEIVR7L2c36sVjQG2TQZQrNnnzOtcgzY1yqj7oqKisj2mQ4cOdOjQ4YnntLKy4tdff81xv6WlJf7+/vj7+2e7f+LEiUyYMIG0tDS1sD+Am5tbjg9v169fx8jIiO7du2Nra/vE+IQQoqBY278eno8V5s+pnRBCvCoy3VKIxzzpoetR95LTctxnaGiIv78/ixYtonTp0rRp0wbIKMD8008/sWzZMqpWrUqjRo0ICAigfPnyTz2foih06dKFM2fOZBmJ9sEHHzBs2DB8fX2pUaMGf/zxB+PGjcvVdQghREFV0tI0T9u9iOTkZAYPHkzJkiUxNTXlnXfe4cSJEwAEBQWp9XHeeustTExMOHz4cJZpjmlpaQwePBgbGxuKFSvGiBEj6NixI+3bt6djx46UKlUKb29vvemWTk5OTJ06ld69e2NpaYmDgwOLFy/Wi2306NG4ublhbm6Os7Mz48aNIzU1+9Ebj/vll18oVqyYunpcprZt29KtW7fnu1lCCAHYWhhTwsL4iW1KWBhj+5Q2QgiRlyRJJsRjcnqYsvt4OrZNMqbqlB2wlL6f+ertP336NBMnTlRf9+3bl+joaNLT0/VWOvv4448JCwsjOTmZuLg4goODadeuHZDxsKPT6fQemh717bffotPp8PPzy7JvxowZ3Lp1i3v37rF27VqGDh3K3bt31f0TJ07k9OnTT78BQghRQNQub4u9tSlKDvsVwN7alNrlX/7oq1GjRrFhwwaWL19OaGgoLi4uNGvWjLi4OLXNmDFjmD59OhEREXqLv2T69ttvWbVqFcuWLSMkJITQ0FB+++03UlNTmTFjRo7nnj17Nl5eXoSFhTFw4EAGDBjAxYsX1f2WlpYEBAQQHh7OvHnzWLJkyVOnJGXq2LEj6enpbN26Vd0WGxvLjh076N27d676EEKInJz4ummOibISFsac+LrpK45ICPGmkySZEI95nR66hBBC5MxAozChtQeA3nu23cfTKfb/X2pMaO2BgSand/RnExUVpY7iStfqOHLlNltOX2f/uWgWLlzIzJkzad68OR4eHixZsgQzMzN+/vln9fhJkybRtGlTKlSokO20yfnz5/Pll1/Srl07KlasyJ49e3BwcMDb25syZcpkaZ+pRYsWDBw4EBcXF0aPHk3x4sU5cOCAuv/rr7+mXr16ODk50bp1a0aMGPHEKaOPMjMz4+OPP2bZsmXqtpUrV6px5bWAgABsbGxeqI+oqCgURZEvhoQoIE583ZTQr5viVrIINmZGuJUsQujXTSVBJoTIF1KTTIjHZD50DVgZigJ6BfwzH7Py8qFLCCHE8/OpYs/CTzzx2xauV0/SztqUCa098Klin+fnDDwfo3e+lNhIUlNTSS3uqrYxMjKidu3aREREUKtWLQC8vLxy7DM+Pp5///2X2rVrq9sMDAx46623nrhyMqA3Kk1RFOzs7IiNjVW3rVu3Dn9/f65cuUJiYiJpaWlYWVnl+nr79etHrVq1uH79OmXKlCEgIICePXuiKPJ3UAiRN2wtjLMtzi+EEK+aJMmEyEZ+PHQJIYR4Pj5V7GnqYcfxyDhi7yVR0jJjtO/L+DIj8HwMA1aGZrsC8tebzlOqdNkc/0YUKVIkz+MBsqx2qSiKmlg7cuQIXbt2xc/Pj2bNmmFtbc3atWuZPXt2rvuvWbMm1atX55dffuH999/nwoUL7NixI0+vQQghhBDidSDTLYXIgU8Vew6Pfo81/d5mXucarOn3NodHvycJMiGEeA0ZaBTqVihGmxplqFuh2EtJkKVrdfhtC8+SIDO0sQcDQ5Kuh+O3LZx0rY7U1FROnDiBh4dHrvq2tramVKlSarF/gPT0dEJDQ18o5j/++ANHR0fGjh2Ll5cXrq6uXLt27Zn76du3LwEBASxbtowmTZpQrly5XB+7fft2bGxsSE9PBzJqeCqKwpgxY/T6/+STT9TXu3btolKlSlhYWODj40NMTIy6T6vVMmnSJMqWLYuJiQk1atQgMDDwiTGcP3+e5s2bY2FhQalSpejWrRu3bt3K9TUIIYQQ4s0gSTIhnuBVPHQJIYQoGI5HxumNLs6kMTbFskYL7hxYytWwENbtPkK/fv148OABffr0yXX/gwYNYtq0aWzZsoWLFy8yZMgQ7ty580LTGl1dXYmOjmbt2rVcuXIFf39/Nm3a9Mz9fPzxx/z9998sWbLkmQv2N2jQgHv37hEWFgZAcHAwxYsX11vUJjg4WK1x9uDBA2bNmsWKFSs4ePAg0dHRjBgxQm07b948Zs+ezaxZszh79izNmjXjgw8+4PLly9me/+7du7z33nvUrFmTkydPEhgYyL///stHH330bDdBCCGEEIWeJMmEEKIQCwoKQlEUvZVOhRDPJ/Ze1gRZpqLePTF3r8+t7bPp2eY9/vzzT3bt2kXRokVz3f/o0aPp0qUL3bt3p27dulhYWNCsWTNMTbNfdTk3PvjgA4YNG4avry81atTgjz/+YNy4cc/cj7W1NR06dMDCwoK2bds+87E1atRQk2JBQUEMGzaMsLAwEhMTuX79On/++SeNGjUCIDU1lR9//BEvLy88PT3x9fVl3759an+zZs1i9OjRdO7cGXd3d7799ltq1KjB3Llzsz3/ggULqFmzJlOnTqVixYrUrFmTpUuXcuDAAS5duvTM90IIIYQQhZei0+myK6tRYCUkJGBtbU18fPwzFaUVQojCKCUlhbi4OEqVKiVFtoV4QUeu3KbLkqNPbbem39vUrVDshc+n1WqpVKkSH330EZMnT37h/l5U48aNqVy5Mv7+/k9tm67V6dWIW7dgCn9evsy2bdsoUaIEhw8fpnPnzkyfPp24uDhGjhzJ9evXCQgI4PPPP+f+/ftqX5s2baJDhw5otVr1c15QUJCaVAMYNmwYZ86cYf/+/URFRVG+fHnCwsKoUaMGHTt2ZMuWLRgbG+vFeP/+fXbu3Enz5s3z7iYJIYQQ4rWU21yRFO4XQohCzNjYGDs7u/wOQ4hCoXZ5W+ytTbkRn5Rt4X6FjAVeape3fa7+r127xu7du2nUqBHJycksWLCAyMhIPv744xeK+0XduXOHoKAggoKC+OGHH57a/vHVPwFMbtty/eAhzpw5g5GRERUrVsTb25ugoCDu3Lmjl/DKbiGCF/lONzExkdatW/Ptt99m2WdvL3VGX0cpKSlZkppCCCHEqyDTLYUQ4jXm5OSUZQpRjRo1mDhxIpDx8PjTTz/Rrl07zM3NcXV1ZevWrWrb7KZbbtiwgcqVK2NiYoKTk1OWVe6cnJyYOnUqvXv3xtLSEgcHBxYvXvyyLlGIAsNAozChdUYh/sfHZWa+ntDa47nrV2o0GgICAqhVqxb169fn3Llz7N27l0qVKj1/0HmgZs2a9OzZk2+//RZ3d/cnts1c/fPx2m0Pbd24n5jIiAlT1YRYZpIsKChIrUf2NFZWVpQuXZqQkBC97SEhITkukuDp6cmFCxdwcnLCxcVF7+dlrTha2D3rYgyHDx+mQYMGmJmZUa5cOQYPHqw3WtDJyYnJkyfTvXt3rKys6N+/f66OE0IIIfKaJMmEEKKA8/Pz46OPPuLs2bO0aNGCrl27EhcXl23bU6dO8dFHH9G5c2fOnTvHxIkTGTduHAEBAXrtZs+ejZeXF2FhYQwcOJABAwZw8eLFV3A1QrzefKrYs/ATT+ys9euE2VmbsvATzxdaAblcuXKEhIQQHx9PQkICf/zxBw0bNnzRkF9YVFQU8fHxesXzs5PT6p8AGlMLjEs4sW/7Bho2zEiSNWzYkNDQUC5duqQ3kuxpRo4cybfffsu6deu4ePEiY8aM4fTp0wwZMiTb9p9//jlxcXF06dKFEydOcOXKFXbt2kWvXr3UJI94Ns+yGMOVK1fw8fGhQ4cOnD17lnXr1nH48GF8fX31+pw1axbVq1cnLCyMcePG5fo4IYQQIi/JdEshhHjNPFrLJzlNi/Yp04x69uxJly5dAJg6dSr+/v4cP34cHx+fLG2/++47GjdurBbudnNzIzw8nJkzZ9KzZ0+1XYsWLRg4cCCQUUx8zpw5HDhw4KmjSIR4E/hUsaeph51eza3a5W3f+BWQc1r9M5NJuSqkxF6lqEtNAGxtbfHw8ODff/99pveWwYMHEx8fz/Dhw4mNjcXDw4OtW7fi6uqabfvMkWejR4/m/fffJzk5GUdHR3x8fNBo5Pvi5/HoYgxeXl7qYgx+fn4kJiYSHx+vLsYwbdo0unbtytChQ4GMFVf9/f1p1KgRCxcuVBemeO+99xg+fLh6jr59++bqOCGEECIvyScDIYR4jQSej+Gdb/fTZclRhqw9zc17yfjvu0zg+Zgcj6lWrZr6e5EiRbCysiI2NjbbthEREdSvX19vW/369bl8+bLeiIpH+1QUBTs7uxz7FOJNZKBRqFuhGG1qlKFuhWJvfIIMnrz6J4Btk/44jt6Oecly6rbTp08TE/Pf+1vPnj2zrMbbtm1bvZpkGo2GCRMm8Pfff5OSksLp06f1vhRwcnJCp9NRo0YNdZurqysbN27kzp07PHjwgIiICObMmSMLmjyjdK2OI1dus+X0dVyr1+bAgSB0Oh2HDh2iffv2VKpUicOHDxMcHEzp0qVxdXXlzJkzBAQEYGFhof40a9YMrVZLZGSk2reXl5feuXJ73LM6fPgwiqKwbt06AObOnYuiKFy7du25+xRCCFF4yEgyIYR4TWTW8nl03JiiKCQ8TGXAylB1KldqaqrecdkVudZqtS8Uy8voUwhRuJW0zN3Inty2E6+XxxdkeBBfgrj9K/hhw94nLsaQmJjIp59+yuDBg7P06eDgoP7+eH243B4nhBBC5CVJkgkhxGsgp1o+GnNr0hMz6ov5bQunTlnzF/oGvVKlStkWvHZzc8PAwOC5+xVCiJe9+qfIP9l9iWNSrjLpyQ8Z7Tedt2rUBjIWY5g+fTp37txRp056enoSHh6Oi4vLM53zeY8TQgghXoRMtxRCiNdATrV8TB2rcf/CAR7+dZ5rf/6Pth99/ELJrOHDh7Nv3z4mT57MpUuXWL58OQsWLHhqQW4hhHial736p8gfOX2JY2BqgVEJJ+5fCOIvk/Kka3XZLsYwevRo/vjjD3x9fTl9+jSXL19my5YtTy3A/7zHAUyZMgUrKysURUGj0VCqVCn279//vLdACCHEG0SSZEII8RrIqZaP9dsfYVKuCrG/TSJ2vR813mlKhQoVnvs8np6e/Prrr6xdu5YqVaowfvx4Jk2apFe0XwghntfLXP1T5I8nLchgWq4K6LQkl6jI8cg4dTEGOzs7dTGGatWqERwczKVLl2jQoAE1a9Zk/PjxlC5d+onnfd7jAOLj4/nss8/Yu3cva9asQVEUWrZsSVpa2rPfACGEEG8URad7yrJpBUxCQgLW1tbEx8djZWWV3+EIIUSuHLlymy5Ljj613Zp+b1O3QrFXEJEQQjy/R1fpldU/C7Ytp68zZO3pp7ab17kGbWqUefkB5SAlTcuKI1Fci3uAo6053eo6YWyYMR7g4sWLVKxYkY0bN1KiRAkaNGjA2rVr6dSpE3PnzmXYsGFERUXh6OiYb/ELIYR4uXKbK5KaZEII8RqQWj5CiMIkc/VPUfAVhAUZpu0MZ8mhSLT//wf0YWQon7aaj+5+HNpHVm4+e/YsjRs3zqcohRBCFAQy3VIIIV4DUstHCCHE6yjzS5yc/voogH0+fokzbWc4iw7+lyADiP1tErqUJKwadKf7+O/ZvHkzAElJ2U8bFUIIITJJkkwIIV4TUstHCCHE6+Z1/hInJU3LkkP6Kz6nxF0HbRrW73yMdZ0OHEpx5Fr0X688NiGEEAWTTLcUQojXiE8Ve5p62EktHyGEEK+NzC9x/LaF6xXxt7M2ZUJrj3z7EmfFkSi9EWQAhjalAIWEE5sxKlaOlNirjDyyJl/iE0IIUfBIkkwIIV4zUstHCCHE6+Z1/BLnWtyDLNs0GkNsGvXg7uGVxK77GgyMeL/nF+z++dt8iFAIIURBI6tbCiGEEEIIIQqcnw9dZfKOiKe2G9eyEn0aOL+CiIQQQryucpsrkppkQgghhBBCiAKnW10nnjaQTaNktBNCCCFyQ5JkQgghhBBCiALH2FBDvwbln9imX4PyGBvKI48QQojckb8YQohXxtvbGzMzM/W1i4sL9vYvVux37ty5KIrCtWvXXjQ8IYQQQhQwX7bw4NOG5bOMKNMo8GnD8nzZwiN/AhNCCFEgSeF+IYQQQgghRIH1ZQsPhr9fkRVHorgW9wBHW3O61XWSEWRCCCGemSTJhBBCCCGEEAWasaFGivMLIYR4YfL1ihDimaSlpdGiRQuMjIxQFAVDQ0OaNm0KQJ06dTA2NkZRFIyMjGjQoAEPHmRdnv1JfTdr1kzt28zMjOHDh+u18fPzU89RtGhRzp8/n6fXJ4QQQgghhBDizSRJMiHEM6lfvz6BgYH07NmTffv24e/vT+nSpQGwtLRk/vz5HDp0iDFjxvDHH3/w4Ycf5rrv5s2bExQUxMSJEzlw4ABdu3blu+++Y968eQAcO3aMiRMnUq1aNXbu3EmXLl1YtmzZS7lOIYQQQgghhBBvFkWn0+nyO4i8lJCQgLW1NfHx8VhZWeV3OEIUeClpWrXGh7X2HiPa16N79+4sX778qce2atWKAwcOcP/+fSCjcP+xY8d4+PAhkFG4//79+8TExKj/7y5atIj+/furfbi7u5OcnExUVBT16tUjNDSUpKQkdf/bb7/NsWPHiIqKwtHRMY+vXgghhBBCCCFEQZfbXJHUJBNC5GjaznCWHIpE+/+p9Htn9wBg7NE42/ZDhgxh2bJl3L9/H61WC4CiKNm2fdyBAwcA+PTTT/n000/19hUpUgSAqKgoHBwc9Pa99957HDt2LHcXJIQQQgghhBBC5ECSZEKIbE3bGc6ig5F62zQm5gBsPfMPzjvD9ZZVX7x4Mf7+/jRp0oRu3bphb2/P9OnT2b9/f67Od+vWLQDmzp1LlSpV9PZZWFi8yKUIIYQQQgghhBBPJUkyIUQWKWlalhyKzLLdzLE6AIln97DkUFWGv19RXV5927ZtGBoasmfPHrW9r69vrs/ZvHlzACIiIhgyZEi2bZycnAgNDdXbljkCTQghhBBCCCGEeBGSJBNCZLHiSJQ6xfJRGlMLTMpV5cGF/dw0NOIruztULJLE4cOHqVatGtu3b2fw4MG0bduW77//nsuXL+f6nKVLl6Z+/fosXryY9PR0unTpQkxMDJs2baJo0aIsWbKE2bNnU69ePWrVqsWkSZPYvn07x48fz8MrF0IIIYQQQgjxppLVLYUQWVyLe5DjvpKdJ2PqWIPEs3uYPbQbn332GTExMXzzzTfUqlWLBQsW0LhxY06cOEGbNm2e6bwHDx6kbdu2/PLLLzRu3Jhu3boRFBREtWrVAKhbty7jxo3jzJkztGjRgpUrV9KjR48XulYhhBBCCCGEEAJkdUshRDZ+PnSVyTsintpuXMtK9Gng/AoiEkIIIYQQQgghnk9uc0UykkwIkUW3uk5onrIopUbJaCeEEEIIIYQQQhQGkiQTQmRhbKihX4PyT2zTr0F5tWi/EEKI/2vvzsOyqvP/j7/Oze5yg7gBpYC7uJuKmCa55FJ+M5fULJcUS8cSl8mswVxqNMtccsrJGnHJ1BYtlyyl3MndNDFTB6UUl0TAJUG4798f/rhHBBSVTc/zcV33lfc5n3PO+9x2BF58FgAAANzrmLgfQLbGdAySJM3ZFJtpEn+LcS0gy9gPAAAAAMD9gDnJANxUappNC6KP6XjCZfl7F9NzIQH0IAMAAAAA3DNymxXRkwzATbk6W5icHwAAAABw36M7CAAAAAAAAEyPkAwAAAAAAACmR0gGmFxaWpo6duwoFxcXGYYhZ2dntW3bVpIUHBwsV1dXGYYhFxcXtWjRQpcvX3YcGxoaKg8PD7344otydnaWYRiqWLGiTp48WVi3AwAAAADAHSEkA0zu4Ycf1po1a9SvXz9FRUVp5syZ8vPzkySVLFlS77//vjZt2qRXX31VW7duVbdu3TIdf+XKFa1YsUKff/653n//fZ04cULdu3cvjFsBAAAAAOCOsbolYDLXr1bpabugUV2aqU+fPpo3b94tj33iiSf0448/6tKlS5Ku9STbsGGDTpw44QjWmjRpopiYGF28eDFf7wMAAAAAgNxgdUsAWUxaHaM5m2Jl+//R+IV9ayVJrkGts20/bNgwzZ07V5cuXZLNZpMkGYaRqY2bm5sjIJMkPz8/7dmzJx+qBwAAAAAg/zDcEjCJSatj9O+N/wvIJMniVkyS9M3PJzVpdUym9h999JFmzpyp4OBgzZ07V99//71atWqlGzuf3hiaGYaRpQ0AAAAAAEUdPckAE0hNs2nOptgs2z3860mSLu5bqzmb6mjkYzXk6nwtO1+xYoWcnZ21du1aR/uhQ4cWTMEAAAAAABQwQjLABBZEH8vUgyyDxb2E3CrU0eUDP+iss4te8zmvGsWvaPPmzapbt65Wrlypl19+WZ07d9a//vUvHT58uOCLBwAAAACgADDcEjCB4wmXc9xXrudEufvX18V9azU1/Dm9+OKLio+P11tvvaXGjRtr1qxZat26tXbs2KEnn3yyAKsGAAAAAKDgsLolYAKfbPqvJq46eMt2EY/X1IAWlQqgIgAAAAAACkZusyJ6kgEm8FxIgCzGzdtYjGvtAAAAAAAwI0IywARcnS0KaxF40zZhLQIdk/YDAAAAAGA2TNwPmMSYjkGSpDmbYjNN4m8xrgVkGfsBAAAAADAj5iQDTCY1zaYF0cd0POGy/L2L6bmQAHqQAQAAAADuW7nNiuhJBpiMq7OFyfkBAAAAALgB3UcAAAAAAABgeoRkAAAAAAAAMD1CMgAAAAAAAJgeIRkAAAAAAABMj5AMAAAAAAAApkdIBgAAAAAAANMjJAMAAAAAAIDpEZIBAAAAAADA9AjJAAAAAAAAYHqEZAAAAMB1Nm/eLMMwtGTJksIuBQAAFCBCMgAAAAAAAJgeIRkAAADuSyNHjpS7u7sMw5DFYpG3t7fOnDkjSerbt6/c3NxkGIbc3NzUs2dPx3EtWrSQJPXs2VOGYcjLy0uSNH36dJUoUUKGYcgwDFmtVm3ZsqXA7wsAAOSPfA3JEhIS1Lt3b1mtVnl5eWnAgAG6ePHiTY8JDQ11fOOR8XrxxRfzs0wAAADcZ/bu3av33ntP7du316ZNm/T555/rsccek81m05AhQ7Rw4UINGzZMGzZs0LBhw7R06VKFhYVJkiIjIyVJU6ZM0c8//6xdu3bpypUrGj58uGrWrKmoqCitWLFCPXr0kMXC75wBALhfOOfnyXv37q34+HitXbtWV69eVf/+/TVo0CAtWrTopseFhYVpwoQJjvfFihXLzzIBAABwnzlw4IAk6e9//7sefvhhSVLXrl0lSXPmzNGQIUM0ZcoUSdIjjzyi3bt3a/HixZozZ44qV64sSapYsaLq1q0rSTp69Kgk6dlnn1WrVq0kSU888UTB3RAAAMh3+farr4MHD2rNmjX6+OOPFRwcrObNm+v999/X4sWLdfLkyZseW6xYMfn4+DheVqs1v8oEAADAfSI1zaZPNv1XY7/+RRd9HlKpUqXUvHlzPfjgg+rTp49iY2N15swZpaWladasWZlGLkRFRenSpUs5nrty5cqqUqWKwsPDVb58eXXp0kV79+4tuJsDAAD5Lt96kkVHR8vLy0uNGjVybGvTpo0sFou2bdump556KsdjP/30Uy1cuFA+Pj7q1KmTIiIicuxNlpKSopSUFMf75OTkvLsJAAAA3BMmrY7RnE2xstn/t80aNk+PXNyh879s0Oeff66FCxdqzpw5kqTBgwc7epZlcHFxuek1Dh8+rM8++0yRkZFau3atli1bpjlz5mjgwIF5fj8AAKDg5VtIdurUKZUrVy7zxZyd5e3trVOnTuV43DPPPCN/f3/5+flp3759Gj16tA4dOqSvvvoq2/aTJk3S+PHj87R2AAAA3DsmrY7RvzfGZt1hWLS3ZLBeGN1Ta9tUkYeHh7755htZLBYdOnRIrVu3zvZ8Gb+cvXr1apZ9vXr1Uq9evSRJJUuW1Pvvv09IBgDAfeK2Q7JXX31Vb7/99k3bHDx48I4LGjRokOPPderUka+vr1q3bq2jR4865oe43pgxYzRixAjH++TkZFWoUOGOrw8AAIB7R2qaTXM2ZQ3ILvz8vS7/uknFg0L1/gk/nfr+hGw2mx566CF5eXlp/vz56tq1qwYNGqSLFy9q1apVOnfunL7++msFBQVJkubNm6f69evLarXq2LFjGj16tPr3769atWpp48aNunTpkqMtAAC49912SDZy5Ej169fvpm0qVaokHx8fxxLbGdLS0pSQkCAfH59cXy84OFiSdOTIkWxDMjc3N7m5ueX6fAAAALh/LIg+lmmIZQan4l5KPX1UV47tlWTXv5xd1K1bN40dO1aSVKJECc2dO1dfffWVDMOQ1WrVCy+8IElyd3dXnz599Omnn6pOnTry9PTU5s2bdfz4cQ0ePFg2m01OTk5q0aKFFixYUHA3CwAA8pVht9uz+bbi7h08eFBBQUHauXOnHnroIUnS999/r/bt2+uPP/6Qn59frs6zZcsWNW/eXD///LNjdaGbSU5Olqenp5KSkpjwHwAA4D439utfND/6+C3b9Qnx14QnaxdARQAAoKjJbVaUb6tb1qxZU+3bt1dYWJi2b9+uLVu2aOjQoerZs6cjIDtx4oRq1Kih7du3S7q2tPbEiRO1a9cuHTt2TN9884369OmjRx55JFcBGQAAAMzF3zv7xZ3utB0AADCvfAvJpGurVNaoUUOtW7dWx44d1bx5c3300UeO/VevXtWhQ4d0+fJlSZKrq6vWrVunxx57TDVq1NDIkSPVtWtXrVixIj/LBAAAwD3quZAAWYybt7EY19oBAADcTL4NtywsDLcEAAAwlxxXt/z/XngkUGM6MsE+AABmldus6LYn7gcAAACKkowAbM6m2EyT+FsMKawFARkAAMgdepIBAADgvpCaZtOC6GM6nnBZ/t7F9FxIgFyd83V2EQAAcA+gJxkAAABMxdXZogEtKhV2GQAA4B7Fr9YAAAAAAABgeoRkAAAAAAAAMD1CMgAAAAAAAJgeIRkAAAAAAABMj5AMAAAAAAAApkdIBgAAAAAAANMjJANgel5eXmrQoEFhlwEAAAAAKESEZAAAAACKrHHjxql+/fqO9/369VPnzp0LrR4AwP2LkAwAAAAAAACmR0gGAJLS09NVt25dGYYhi8WiFi1ayGazSZKSk5PVqFEjOTk5yTAMlShRQtOnTy/cggGYUn4PD7/x/H/++aceeOABGYYhwzB0/PjxfLs2AABAYSMkAwBJ+/fvl5OTk1avXq0XX3xRmzdvVr9+/SRJTZo00W+//aYZM2YoKipKoaGhGj58uNauXVu4RQMwnV27dmnVqlWSJGdnZz311FP5er2XXnpJ8fHx+uqrr/Tzzz+rQoUK+Xo9FE0BAQFZfjlUv359jRs3Tna7XePGjVPFihXl5uYmPz8/vfzyy452hmFo+fLlmY718vJSZGSk4/3o0aNVrVo1FStWTJUqVVJERISuXr2aq9rmz5+v0qVLKyUlJdP2zp0767nnnrut+wQAwLmwCwCAwpCaZtOC6GM6nnBZqWk2ubq6ateuXbJYLOrQoYN2796tJUuWaPDgwTp06JB27dqlhg0bSpJatWolb29vvfHGG2rbtm0h3wkAM6lcuXKBXu/w4cOyWq35Hsbh3vXll19q2rRpWrx4sWrVqqVTp07p559/vq1zlCxZUpGRkfLz89P+/fsVFhamkiVL6pVXXrnlsd27d9fLL7+sb775Rt27d5cknTlzRqtWrdL3339/R/cEADAvepIBMJ1Jq2NUI+JbTVx1UPOjj+tKWrpsxUvr7TW/Otq0atVKqamp+u677yRJDz30kGO4kWEYOn/+vP7444/CugUAJpUxHNLLy0vp6elavny5498lSdqyZYvKly8vi8UiwzDk7u6u8ePHO45ftmyZypYt6xhaXqlSJR06dCjHa+3atUtJSUkyDENeXl4FcYsoItJtdkUfPaev955QSppNNrs923ZxcXHy8fFRmzZtVLFiRTVp0kRhYWG3da1//OMfatasmQICAtSpUyeNGjVKS5cuzdWxHh4eeuaZZzR37lzHtoULF6pixYoKDQ29rToAAKAnGQBTmbQ6Rv/eGJvtvoztYzoGObadP39ekrR69Wq5urpmal+2bNl8qhIAbm7Xrl2qXr26WrZsqWnTpjm2d+nSRWlpafr8889VunRp/fDDD/L29pYkHT9+XF27dlXTpk01b948JSUlaejQoWratKnj37obr/HEE0/ojz/+0JYtW1S8ePECuz8UrjW/xGv8ihjFJ12RJJ29kKKZUYcV1DZe7Wv7ZmrbvXt3TZ8+XZUqVVL79u3VsWNHderUSc7Ouf8xY8mSJZo5c6aOHj2qixcvKi0tTVarNdfHh4WFqXHjxjpx4oQeeOABRUZGql+/fo7wGACA3CIkA2AaqWk2zdmUfUCWlvynJGnOpliNfKyGfvzxR7m6uqpjx46Ob9yHDh1akOUCQI4yhl1arVbVrVvXsT05OVlNmzZV165dJSlTT5oBAwbI29tbW7dudWyrUqWKmjRpou+++07t2rXLcg13d3c5OTllugbub2t+idfghbt1fb8xwzCU/NdVDV64Wx8+21Dta/s65gyrUKGCDh06pHXr1mnt2rUaMmSI3nnnHW3YsEEuLi4yDEP2G3qhXT/fWHR0tHr37q3x48erXbt28vT01OLFizV16tRc19ygQQPVq1dP8+fP12OPPaYDBw445u4DAOB2EJIBMI0F0cdky360iJR+VScjh6lUi+fUvscS/fTTT3rmmWfUrl07BQQEaPjw4YqLi9MTTzyho0ePavHixQoJCdG4ceMK8hYAmNCNcyjmMOpNkvT0009r/vz5slqtatCggV566SV169ZNkvTrr7/q3Llz2fau2bZtW5aQDOaTbrNr/IoY3fi/mKWYp9IvJkiSxq+IUfCDxRQb+79fOnl4eKhTp07q1KmT/va3v6lGjRrav3+/GjZsqLJlyyo+Pt7R9vDhw7p8+bLj/datW+Xv76/XX3/dse1OVlEdOHCgpk+frhMnTqhNmzYsMgEAuCOEZABM43jC5Rz3uZTxlz0tVWe+GKczkpo1a6YFCxZIkg4cOKAOHTpo2rRpeuedd2SxWFS+fHkNGjSogCoHYFaTVsdozqZYR8B/JS1dB+OTNWl1TLbt582bpyFDhmjGjBlav369unfvrq5du+qLL75QSkqK/Pz8NH/+/CzH1alTJz9vA/eI7bEJjiGW13P3r6tL+6PkUaWJjp8toc5PvycnJydJUmRkpNLT0xUcHKxixYpp4cKF8vDwkL+/v6Rrc3zOmjVLISEhSk9P1+jRo+Xi4uI4d9WqVRUXF6fFixercePGWrVqlZYtW3bbtT/zzDMaNWqU5syZk+3/4wAA5AYhGQDT8Pculu32iuGZJweOeLymBrSo5HhfrFgxbdiwIV9rA4Ab5TSHol3X5lC0y1BaWlqW/cHBwVq0aJEkKSQkxDHsrFq1atq5c6cefvhhubu752vtuDeduZA1IJMkz6ZPKy3xtM58MUEWt+LqPuo1JZ05IenaAg+TJ0/WiBEjlJ6erjp16mjFihUqXbq0JGnq1Knq37+/WrRoIT8/P82YMUO7du1ynPv//u//NHz4cA0dOlQpKSl6/PHHFRERcds9tT09PdW1a1etWrVKnTt3vqP7BwDAsN84ScA9Ljk5WZ6enkpKSrqtCT8B3P9S02yqEfFtzkMuJVkM6deJHeTqzOK/ADLLGKZYEN865fTvVdz0p+Xs6SO//jP1+4xeKutZTCtXrlTJkiVVvXp1NWjQQN26dVOLFi30+++/64UXXpC3t7fi4uK0e/duNWrUSH5+fnrrrbdUoUIFRUdHKzIyUgcOHJCrq6u8vLwUGBioPXv2SLo211NsbKwSExPz/Z5R+KKPnlOvOT/dst1nYU0VUrl0AVR0e1q3bq1atWpp5syZhV0KAKCIyW1WRE8yAKbh6mxRWIvAHFe3lKSwFoEEZACyNX78eD3wwAMFcq2bzqH4/3m17KeEHz9S48aNJV0L79LT0zVu3DilpaXJMAwFBgZqzZo1kqSGDRvqu+++U9++fdW/f3/Z7Xa5urqqVq1at7USIe5fTQK95evprlNJV7LMSyZJhiQfT3c1CfQu6NJu6vz581q/fr3Wr1+vDz74oLDLAQDcw+hJBsB0bpzjR7rWgyysRaDGdAwqvMIA4P8b+/Uvmh9968nL+4T4a8KTtQugIphFxuqWkjKvcPn//5uxumVREhAQoPPnzysiIkKjRo0q7HIAAEUQPckAIAdjOgZp5GM1HKvF+XsX03MhAfQgA3BT1w+3tFgsWYZd5uXvHXOaQ/FO2wG51b62rz58tqHGr4jJNIm/j6e73ugUVOQCMkk6duxYYZcAALhP0JMMAAAgFzJCMn9/fx0/flzOzs7q0KGDDh48qCNHjuRpSMYciihs6Ta7tscm6MyFKypX8toQSyeLcesDAQAoguhJBgAAkA/i4+MlSbNnz9aAAQPy5RrMoYjC5mQxiuTk/AAA5Ce+swIAAMjBiYS/VHfcGlUes8qxLWNi8IEDB8owDDk7O2vZsmV5fu0xHYP0wiOBurHzjsWQXniEORQBAADyGj3JAAAAslHt9dVKTc863vHt//rq8OHDatiwoS5cuKD09HR16dJF77zzTp5PGs4cigAAAAWH77AAAABukFNAJkmp6XZ1nPubkpOTZbfbtXjxYknS66+/ni+1uDpbNKBFJU14srYGtKhEQIb7wubNm2UYhpYsWVLYpQAA4EBPMgAAgOucSPgrx4BMko5P7SKlparS4iqqXaumVq5cKUmqWbNmQZUIAACAfEBIBgAAcJ0OMzfcdL/h4i57Wqpi/3tEsf89IklycXHR3r17C6A6AAAA5Bf66wMAAFznUkp6ttv9R6+U/+iVqvjyIvmPXqlKr66U3W6X3W5XampqAVcJFH1vvvmmrFarDMOQxWJR+fLl9cMPP+TYfsaMGSpRooQMw5CTk5OaNm2qK1euOPZ7eXmpXr16atKkiSwWi5ycnBQaGprpHKtXr3Zc083NTVOmTJFhGBozZkx+3SYA4D5CSAYAAHCd4m5OedoOMKukpCS9+OKLWrdunT777DMZhqHHH39caWlpWdru3LlT4eHhCgwM1KpVq/TKK69o+/btat++faZ2+/btU/HixfXdd99pwIAB2rBhgyZPnixJSk1N1VNPPSUXFxctWbJE06ZN04QJEwrkXgEA9weGWwIAAFzn25db6uEpOfd2ub4dgP9JTbNlWon1rUlvZ1poon79+qpRo4ZWrFihsmXLZjp22LBhcnV11c8//yyLxaKOHTsqNjZWS5YsUVpampydr/3YYrVa9eOPP0qS2rZtq0WLFmn58uV69dVXNWXKFKWmpurHH39U3bp1JUkXL17U6NGjC+gTAADc6+hJBgAAcJ0HvD3k6mTctI2rk6EHvD0KqCKg6Ju0OkY1Ir7VxFUHNT/6uCauOqiKz74lr3J+cnFxkWEYqlGjhqRrvcFuFBsbq4oVK8pi+d+PJ126dJEk7dixw7GtYsWKmY4rWbKkEhISJEm7d++Ws7OzIyCTpGeeeSbvbhIAcN8jJAMAALjBb291zDEoc3Uy9NtbHQu4IqDomrQ6Rv/eGCvbDYvCnv58gi5cuKg2fYZpxYoVWr58uSRlmmfsdmX0KLuezWa74/MBAHA9QjIAAIBs/PZWR215pZWs7k5yMiSru5O2vNKKgAy4TmqaTXM2xWbdnnBCsqXJs/kzOlTuUT3WvqNOnDiR43kCAwMVFxeXKfD66quvJEmNGzfOVS0NGzZUWlqafvnlF8e2zz77LLe3AgAAc5IBAADk5AFvD+0b1/7WDQGTWhB9LEsPMkly9iovyVDyjuVyKV1Bzw7brNXz3s/xPDNmzFDjxo1Vv359TZ48WZs3b9bSpUvVsmXLbHuPZeeVV17RxIkTFRoaqtmzZ+vs2bOOifsN4+ZDqAEAkOhJBgAAAOAOHU+4nO12i8VZXi37Kv3iOZ1Z8g8tm/Oe3nrrrRzP06hRI02fPl3//e9/9fjjj+vtt99WkyZNtGbNmlzX4urqqmXLlik1NVXdu3dXeHi4hg8fLkkqXrz47d0YAMCUDLvdns3vfu5dycnJ8vT0VFJSkqxWa2GXAwAAANy3Ptn0X01cdfCW7SIer6kBLSoVQEWZffjhhxoyZIiioqLUqlWrAr8+AKBoyG1WxHBLAAAAAHfkuZAAvbX6YLZDLjNYjGvt8lO/fv2UmJio6tWrq1SpUmrevLm2bt2qsWPHqmTJkrkKyNavX69HH31U58+fl5eXlyIjIxUeHq7ExMR8rR0AUHQQkgEAAAC4I67OFoW1CNS/N2advD9DWItAuTrn7ywvM2bMkN1u1yuvvKJp06bp6tWrslgsCgwM1LfffpurczRr1kzx8fHy9PTM11oBAEUXc5IBAAAAuGNjOgbphUcCZblhbnyLIb3wSKDGdAzK9xo8PT3l5eWljz76SKmpqbLb7UpPT9eRI0dUtWrVXJ3D1dVVPj4+eTrJf2pqap6dCwCQ/wjJAAAAANyVMR2D9OvEDop4vKb6hPgr4vGa+nVihzwPyL744gvVqVNHHh4eKl26tNq0aaNLly6pX79+6ty5s6NdaGioXnrpJYWHh6tUqVIqX7685syZo0uXLql///4qWbKkqlSpkqmX2fr162UYRo7DK48ePaonn3xS5cuXV4kSJdS4cWOtW7cuU5uAgABNnDhRffr0kdVq1aBBg7KcxzAMLV++PC8+DgBAHiMkAwAAAHDXXJ0tGtCikiY8WVsDWlTK8yGW8fHx6tWrl55//nkdPHhQ69evV5cuXZTTOmTz5s1TmTJltH37dr300ksaPHiwunfvrmbNmmn37t167LHH9Nxzz+ny5exX6LzRxYsX1bFjR0VFRWnPnj1q3769OnXqpLi4uEzt3n33XdWrV0979uxRRETEXd83AKDgEJIBAAAAKHJsNpumTJmiKlWqyM3NTQ0aNFBaWpq6dOmiCxcuaNiwYRo5cqT8/f21detWpaWlOY799ddf5eLiIovFoubNm+vdd9+VxWKRt7e3fvvtNwUHB2v58uU6d+6c9u3bJ0k6deqUJOnLL79Us2bNFBYWpgsXLmjDhg2SpHr16mngwIGaNm2aHnvsMb377ruy2WwaMWJEprpbtWqlUqVK6f/+7/8UFBQkX19fDR06VNK1nmaS9NRTT8kwDMf7cePGqX79+lqwYIECAgLk6empnj176sKFC5k+j0mTJikwMFAeHh6qV6+evvjiC8f+8+fPq3fv3ipbtqw8PDxUtWpVzZ07V9K1YZ9Dhw6Vr6+v3N3d5e/vr0mTJuXh3xYA3B8IyQAAAAAUOWPGjNHkyZPV64XhmvHFDxr73hzVqFFDtWvXVpMmTZSYmKh169bp888/V3x8vCPsynDhwgWdPHlSGzdu1HvvvaerV69qx44dKlWqlLZt26bBgwdLkmJiYjIdN3bsWI0cOVLjx4+Xk5OTOnXqpHPnzunixYsaNWqUvv76a509e1aGYejq1av65ptvtHTpUsfx6enp+tvf/qZBgwZp//79+uabb1SlShVJ0o4dOyRJc+fOVXx8vOO9dG045/Lly7Vy5UqtXLlSGzZs0OTJkx37J02apPnz52v27Nk6cOCAhg8frmeffdYR4kVERCgmJkbffvutDh48qA8//FBlypSRJM2cOdNR56FDh/Tpp586AjoAwP+wuiUAAACAQpdus2t7bILOXLii4sZVTZs+QxUe/5sWnAuQtiRKssjnufdV+8gyrfzsP0pNTdVTTz2lbdu2KTg4WFFRUTp9+rTKly8vSXJ3d9fMmTNlsVhUvXp1vfDCC0pLS9Nrr70mSXrttdc0duxYHThwIFMdYWFh6tq1qyIjI+Xh4SGr1apPPvlE//3vf7V27VrNmTNHVapUkYeHh7p16ya73a6lS5fq6aefliRt3LhRI0eO1LBhwxznbNy4sSSpbNmykiQvLy/5+Phkuq7NZlNkZKRKliwpSXruuecUFRWlt956SykpKfrnP/+pdevWKSQkRJJUqVIlbd68Wf/+97/VsmVLxcXFqUGDBmrUqJEkZQrB4uLiVLVqVTVv3lyGYcjf3//u/8IA4D5ESAYAAACgUK35JV7jV8QoPumKJCnl5CFdTU3RlXJBcrmu3enkFB3cf0LV6jTU7uiN8vf317Jly1SuXDlJ0qFDhxwhWenSpWWx/G/gjJOTk3x9fTO9l6SkpKRMtWQEWtK1SfYbNWqkgwcPaufOnerXr59OnjypN998U8ePH9e5c+dksVjk5uYm6VovsqSkJLVu3fq2P4OAgABHQCZJvr6+OnPmjCTpyJEjunz5stq2bZvpmNTUVDVo0ECSNHjwYHXt2tUx31rnzp3VrFkzSVK/fv3Utm1bVa9eXe3bt9cTTzyhxx577LZrBID7HcMtAQAAABSaNb/Ea/DC3Y6ATJIMF7cs7VJOHlJi9FKlX0rUf0+d1xdffKmzZ8+qZs2a2Z73+oAsQ0Ywdr2cJv6/UdWqVfXJJ59o5MiRateunWrVqqXixYsrKChIqamp1+o2jFydKzsuLi6Z3huGIZvNJunaogGStGrVKu3du9fxiomJccxL1qFDBx0/flzDhw/XyZMn1bp1a40aNUqS1LBhQ8XGxmrixIn666+/9PTTT6tbt253XOu94sZVT7MTGhqq8PDwAqkHQNFHSAYAAACgUKTb7Bq/IkY3xlQupfxkOLvpyvGfHdssrsV05fdfdPnodl38/aBGjX5VU6dOVYcOHRw9rqpXr37XNe3cudPxZ7vdrl27dqlmzZp67733lJaWpqtXr2rhwoV6+umn1ahRo0w90TIWB4iKisrx/C4uLkpPT7+tmoKCguTm5qa4uDhVqVIl06tChQqOdmXLllXfvn21cOFCTZ8+XR999JFjn9VqVY8ePTRnzhwtWbJEX375pRISEm6rjnvNjBkzFBkZWdhlALiHMNwSAAAAQKHYHpuQqQdZBsPZVdbgrkpcP1eGk7PcHgiSLeWSilVvrrJPvaaTHw1ShSo1FRoaqh9//FGxsbHq27evY6hl+/btlZiYmOmcTZs2Vf369TNt8/f3V506dRz7Jek///mP6tatq6ZNm6pnz55atGiRnn/+eZUpU0avvPKKIiIiNGfOHAUGBurUqVPas2ePvL29JUnHjh3TvHnz9OKLL6pcuXLq0KGDLly4oC1btuill16SdG1YZVRUlB5++GG5ubmpVKlSt/ycSpYsqVGjRmn48OGy2Wxq3ry5kpKStGXLFlmtVvXt21djx47VQw89pFq1aiklJUUrV6509LJ777335OvrqwYNGshisejzzz+Xj4+PvLy8cv13VZBSU1Pl6up61+fx9PTMg2oAmAk9yQAAAAAUijMXsgZkGTwf7ilr46eUuOlTnfx4sM5+87ZslxNlcXFXuacnKOVSsho3bqxu3bqpdevWmjVrVp7UNHnyZE2ePFn16tXT5s2b9c033zhWiXzhhRfUpUsX9ejRQ8HBwTp37pyGDBmS6fi+fftq+vTp+uCDD1SrVi098cQTOnz4sGP/1KlTtXbtWlWoUMExn1huTJw4UREREZo0aZJq1qyp9u3ba9WqVQoMDJQkubq6asyYMapbt64eeeQROTk5afHixZKuhWxTpkxRo0aN1LhxYx07dkyrV6/OdkhqfggNDdXQoUM1dOhQeXp6qkyZMoqIiHAMdQ0ICNDEiRPVp08fWa1WDRo0SJL05ZdfqlatWnJzc1NAQICmTp3qOOdrr72m4ODgLNeqV6+eJkyYICnrcMtLly6pT58+KlGihHx9fTOdL0NKSopGjRqlBx54QMWLF1dwcLDWr1+fh58GgKLMsOd2EP49Ijk5WZ6enkpKSpLVai3scgAAAADkIProOfWa89NtHWNI8vF01+bRreRkufM5wG507NgxBQYGas+ePVl6nOHuhIaGateuXRowYIAGDx6snTt3atCgQZo+fbrCwsIUEBCg8+fPa+zYsY5QKzExUU2aNNG4cePUo0cPbd26VUOGDNEHH3ygfv366cCBA6pdu7aOHDmiypUrS5Jj2+HDh1WlShX169dPiYmJWr58uSRpyJAhWrVqlf7zn/+oXLlyeu2117RhwwY9//zzmj59uqRrq5vGxMRo8uTJ8vPz07Jly/SPf/xD+/fvV9WqVQvh0wOQF3KbFTHcEgAAAEChaBLoLV9Pd51KupJlXrLsZERib3QKytOADHkr3WbX9tgEnblwReVKussuqUKFCpo2bZoMw1D16tW1f/9+TZs2TWFhYZKkVq1aaeTIkY5z9O7dW61bt1ZERIQkqVq1aoqJidE777yjfv36qVatWqpXr54WLVrkaPPpp58qODhYVapUyVLTxYsX9cknn2jhwoWO1UfnzZunBx980NEmLi5Oc+fOVVxcnPz8/CRJo0aN0po1azR37lz985//zJfPC0DRwXBLAAAAAIXCyWLojU5Bkv4XgN2Mj6e7Pny2odrX9s3fwnDH1vwSr+Zv/6Bec37SsMV71WvOT9oTd14PVq+bafXPkJAQHT582LGIQaNGjTKd5+DBg3r44YczbXv44YczHdO7d28tWrRI0rVFFj777DP17t0727qOHj2q1NTUTEM0vb29My32sH//fqWnp6tatWoqUaKE47VhwwYdPXr0Lj4VAPcKepIBAAAAKDTta/vqw2cbavyKmEyT+Pt6uivi8SCVKu7q6JHUJNA733qQBQQE6D6biabArfklXoMX7s7SKzA1zaboo+e05pf4HAPO4sWL3/b1evXqpdGjR2v37t3666+/9Pvvv6tHjx53UPk1Fy9elJOTk3bt2iUnJ6dM+0qUKHHH5wVw7yAkAwAAAFCo2tf2Vdsgn0xD9PIzEEPeS7fZNX5FTI7DZlNO/qbxK2LUNshHThZDP/30k6pWrZoljMpQs2ZNbdmyJdO2LVu2qFq1ao5jHnzwQbVs2VKffvqp/vrrL7Vt21blypXL9nyVK1eWi4uLtm3bpooVK0qSzp8/r99++00tW7aUJDVo0EDp6ek6c+aMWrRocQefAoB7HSEZAAAAgELnZDEUUrl0YZeBO7Q9NiFTT8AbpV04qwNfva/PGxWX/c9Yvf/++9muLplh5MiRaty4sSZOnKgePXooOjpas2bN0gcffJCpXe/evfXGG28oNTVV06ZNy/F8JUqU0IABA/T3v/9dpUuXVrly5fT6669nWuGzWrVq6t27t/r06aOpU6eqQYMGOnv2rKKiolS3bl09/vjjt/GJALgXEZIBAAAAAO7KmQs5B2SSVLxWK9nTUjWgy2Nyc3HWsGHDNGjQoBzbN2zYUEuXLtXYsWM1ceJE+fr6asKECerXr1+mdt26ddPQoUPl5OTkWBkzJ++8844uXryoTp06qWTJkho5cqSSkpIytZk7d67efPNNjRw5UidOnFCZMmXUtGlTPfHEEzc9N4D7g2G/zwbe53ZZTwAAAABA3og+ek695vyU7b5Ti16Va7lK8m4zSJ+FNaXHIIACl9usiNUtAQAAAAB3pUmgt3w93W+6Sqmv57W55gCgqCIkAwAAAADcFSeLoTc6BUlSjkHZG52CWIwBQJFGSAYAAAAAuGvta/vqw2cbysfTPdP2BoOna2nkbLWv7VtIlQFA7jBxPwAAAAAgT7Sv7au2QT7aHpugMxeuqFzJa0Ms6UEG4F5ASAYAAAAAyDNOFoPJ+QHckxhuCQAAAAAAANMjJAMAAAAAAIDpEZIBAAAAAADA9AjJAAAAAAAAYHqEZAAAAAAAADA9QjIAAAAAAACYHiEZAAAAAAAATI+QDAAAAAAAAKZHSAYAAAAAAADTIyQDAAAAAACA6RGSAQAAAAAAwPQIyQAAAAAAAGB6hGQAAAAAAAAwPUIyAAAAAAAAmB4hGQAAAAAAAEyPkAwAAAAAAACmR0gGAAAAAAAA0yMkAwAAAAAAgOkRkgEAAAAAAMD0CMkAAAAAAABgeoRkAAAAAAAAMD1CMgAAAAAAAJgeIRkAAAAAAABMj5AMAAAAAAAApkdIBgAAAAAAANMjJAMAAACQa5cvX1bXrl1ltVplGIYSExOz3RYQEKDp06fn6pyRkZHy8vLK17ozhIaGKjw8vECuBQC4txCSAQAAAMi1efPmadOmTdq6davi4+Pl6emZ7bYdO3Zo0KBBuTpnjx499Ntvv+VpnevXr3cEdgAA5IZzYRcAAAAA4N5x9OhR1axZU7Vr177ptrJly+b6nB4eHvLw8MhV26SkJHl6eua+YAAAcomeZAAAAIDJfPHFF6pTp448PDxUunRptWnTRpcuXcp2KGLnzp3Vr18/SdeGKk6dOlUbN26UYRgKDQ3NdpukLMMtExMT9cILL6h8+fJyd3dX7dq1tXLlSvXt21cWi0WGYcgwDLm5uWnhwoX6+uuvFRgYKMMwVKxYMUebTp06SZKjfcbL3d1d33zzjSTp2LFjevTRRyVJpUqVkmEYjnu43oQJEzIFexnq16+viIiIu/uQAQD3HEIyAAAAwETi4+PVq1cvPf/88zp48KDWr1+vLl26yG633/LYr776SmFhYQoJCVF8fLy++uqrbLfdyGazqUOHDtqyZYsWLlyomJgYTZ48WU5OTkpMTFS9evXk4eGhyZMnS5L69OmjZ599Vh07dpQk/fXXX3J3d1fXrl0VERGhNm3aSJLc3d01dOhQtW7dWikpKerSpYsSEhJUoUIFffnll5KkQ4cOKT4+XjNmzMhSV8ZnsGPHDse2PXv2aN++ferfv//tf7gAgHsaIRkAAABwn0u32RV99Jy+3ntC320/qLS0NHXp0kUBAQGqU6eOhgwZohIlStzyPN7e3ipWrJhcXV3l4+Mjb2/vbLfdaN26ddq+fbu++uortW3bVr4P+mtbmr8WnfRW3f5vavDfXpKrq6tGjx6tDRs2yG63q2XLlo5eaXXq1NFHH32krVu3qm3btoqKipIkjRo1Su+//77WrVungIAApaena/v27XJycnLUUa5cOfn4+GQ7RPPBBx9Uu3btNHfuXMe2uXPnqmXLlqpUqdKdfNQAgHsYc5IBAAAA97E1v8Rr/IoYxSddkSTZbemyVm6gmrVq6/EO7fXYY4+pW7duKlWqVL7VsHfvXj344IOqVq2awubv0NqYM459q79cq4Tv/iXZ0mQYhmP7t99+q7Vr10qSDh48qLCwMF25ckUxMTGONm+++abefPPNTNc6c+aMbkdYWJief/55vffee7JYLFq0aJGmTZt2J7cJALjH0ZMMAAAAuE+t+SVegxfudgRkkmRYnFSq6wR5dh4r1zIV9P7776t69eqKjY2VxWLJMuzy6tWrd11HxqT8NwZkkpTw7QzJbpMsTgrp84pmzpwpSapZs6beffddSVJ4eLj279+vw4cP6/Tp045jX3rpJc2bN8/x8vDwkM1mu63aOnXqJDc3Ny1btkwrVqzQ1atX1a1bt7u5XQDAPYqeZAAAAMB9KN1m1/gVMcp2pjHDkPuDQfqvZ0PtnDlFlQIDtGzZMpUtW1bx8fH/O0d6un755RfHJPh3qm7duvrjjz+0evMeuXg/4Nh+5eQhSZJbxTpKPXVEJ30f0eEj13qPXbx4UT4+PpKkMmXKqEqVKpLk+K8ktWrVSp07d3a8f/nllx1/dnV1ddzDzTg7O6tv376aO3euXF1d1bNnz1yvtAkAuL8QkgEAAAD3oe2xCZl6kGVIOXlIV47/LPeABvo92VPvzJ6vs2fPqmbNmipevLhGjBihVatWqXLlynrvvfeUmJh417W0bNlSFYIe0sll/1SpVgPlXMpXaef+ULrtWoCV8keMZBj687tZmvXzd5KkuLg4LV26VJJ0+vRpLV68WL/88ovefPNNhYSEKDo6WuHh4Tp37pySkpK0bNkyXbhwwXFNf39/GYahlStXqmPHjvLw8Mhx3rWBAweqZs2akqQtW7bc9f0CAO5NhGQAgEJ17NgxBQYGas+ePapfv77Wr1+vRx99VOfPn5eXl1dhlwcA96wzF7IGZJJkcS2mK7//ouSdX8uWclkfPFhBU6dOVYcOHXT16lX9/PPP6tOnj5ydnTV8+PC77kWWocnAN7X6k3f054p3ZL96Rc5evvIK7afidR/TpX3fS5Iu7V2jaq266bcfvlCzZs20b98+SdIHH3ygLVu2aODAgZKkrVu3yjAMxcXFObY5OTnJYvnfbDIPPPCAxo8fr1dffVX9+/dXnz59FBkZmW1tVatWVbNmzZSQkKDg4OA8uV8AwL3HsOdmred7SHJysjw9PZWUlCSr1VrY5QAAboGQDADyR/TRc+o156dbtvssrKlCKpfO93oilu/Xgp/ibtnuuaYVNbFznXyv53p2u11Vq1bVkCFDNGLEiAK9NgAg/+U2K2LifgAAAOA+1CTQW76e7jJy2G9I8vV0V5NA7wKp57WOQXnaLq+cPXtWs2bN0qlTp9S/f/8CvTYAoGghJAMA5Ls1a9aoefPm8vLyUunSpfXEE0/o6NGjuT7+yy+/VK1ateTm5qaAgABNnTrVsW/WrFmqXbu24/3y5ctlGIZmz57t2NamTRv94x//yJubAYB7hJPF0BudrgVONwZlGe/f6BQkJ0tOMVre8nB1Utugcjdt0zaonDxcnQqkngzlypXThAkT9NFHH6lUqVIFem0AQNFCSAYAyHeXLl3SiBEjtHPnTkVFRcliseipp56SzWa75bG7du3S008/rZ49e2r//v0aN26cIiIiHPPKtGzZUjExMTp79qwkacOGDSpTpozWr18vSbp69aqio6MVGhqaT3cHAEVX+9q++vDZhvLxdM+03cfTXR8+21Dta/sWaD1z+jTOMShrG1ROc/o0LtB6pGtDLc+ePatnnnmmwK8NAChamJMMAJDn0m12bY9N0JkLV1Su5LWhPNf3VPjzzz9VtmxZ7d+/XyVKlLjpnGS9e/fW2bNn9f333zuOf+WVV7Rq1SodOHBAdrtdZcuW1ezZs9WtWzc1aNBAPXr00IwZMxQfH68tW7bo0UcfVWJioooVK1YYHwcAFLpb/btc0P5KTdc/V8fo2LnLCihdTK91DCrwHmQAAPPIbVbE6pYAgDy15pd4jV8Ro/ik/62q5nX1T3kdXKa4X/fpzz//dPQgi4uLU1DQzeeeOXjwoJ588slM2x5++GFNnz5d6enpcnJy0iOPPKL169erTZs2iomJ0ZAhQzRlyhT9+uuv2rBhgxo3bkxABsDUnCxGgUzOn1serk4FPjk/AAC3QkgGAMgza36J1+CFu3VjF+UDkf+Qs7Wsxr4+WZ0friObzabatWsrNTU1T64bGhqqjz76SJs2bVKDBg1ktVodwdmGDRvUsmXLPLkOAAAAgPsXc5IBAPJEus2u8StisgRk6X8lKy3hD3k166FvzpZWteo1dP78+Vyft2bNmtqyZUumbVu2bFG1atXk5HRtaE7GvGSff/65Y+6x0NBQrVu3Tlu2bGE+MgAAAAC3RE8yAECe2B6bkGmIZQaLewlZPKy68PN3iivhrQ8WJWvBzEm5Pu/IkSPVuHFjTZw4UT169FB0dLRmzZqlDz74wNGmbt26KlWqlBYtWqSVK1dKuhaSjRo1SoZh6OGHH777GwQAAABwX6MnGQAgT5y5kDUgkyTDsKjM/72i1FNHdPKTv+m9Ca/rnXfeyfV5GzZsqKVLl2rx4sWqXbu2xo4dqwkTJqhfv37XXcNQixYtZBiGmjdvLulacGa1WtWoUSMVL178ru4NAAAAwP2P1S0BAHki+ug59Zrz0y3bfRbWtEhNHg0AAADg/pbbrIieZACAPNEk0Fu+nu4ycthvSPL1dFeTQO+CLAsAAAAAcoWQDACQJ5wsht7oFCRJWYKyjPdvdAqSkyWnGA0AAAAACg8hGQAgz7Sv7asPn20oH0/3TNt9PN314bMN1b62byFVBgAAAAA3x+qWAIA81b62r9oG+Wh7bILOXLiiciWvDbGkBxkAAACAooyQDACQ55wsBpPzAwAAALinMNwSAAAAAAAApkdIBgAAAAAAANMjJAMAAAAAAIDpEZIBAAAAAADA9AjJAAAAAAAAYHqEZAAAAAAAADA9QjIAAAAAAACYHiEZAAAAAAAATI+QDAAAAAAAAKZHSAYAAAAAAADTIyQDAABAngoNDZVhGPL39y/sUgAAAHKNkAwAAAAAAACmR0gGAACAbLm4uMgwjEyvefPmSZIsFkuWfRk2bNggSYqLi8uyDwAAoKgiJAMAAEAWvXr1UlpamiSpZcuWqlSpkiQpNTVVFotFdrtdzs7OatGihZydnSVdC84kqXz58pKkYsWK6dlnn9Xo0aML4Q4AAABuj3NhFwAAAICiZ+PGjZKk1q1ba926dZn2DRo0SBaLRVevXnVsMwxDdrtdklSjRg2dPn1aZcqU0YIFCwquaAAAgLtATzIAAAA4xP15WUER3yqtwwRJUlRUlGPI5BtvvOEYbmmz2bIdagkAAHCvoicZAAAAJElVXlulNNu1P3uU8VHZl5bq7L/7S6mXJEkTJkxQ2bJlJV3rOdaoUaNMx7u7uxdovQAAAHmJkAwAAACZArIMxYoVk//wJZKk1IRTip8zUGfPnpUk2e12bd++PdtzZQRp1w/HBAAAKOoIyQAAAEwu7s/LWQKy49OfkVKSZRTzkkvZAKXG7ZMkubm5KTU1VXa7XYZhqH79+kpOTlZsbKzsdrvsdrsmTpyoL774QvHx8erfv7+qV6+uV199tRDuDAAAIPcMe8YMq/eJ5ORkeXp6KikpSVartbDLAQAAKPKCIr7V5auZU7I/Phqs9PO/Z2mb8a2jk5OTbLbMxzg7Ozt6j2WsgHnjcQAAAAUtt1kRPckAAABM7q8bAjJJenDQh1m2XT89f3p6+k3PeWOABgAAUNSxuiUAAIDJebjk7lvC3LYDAAC4F/GdDgAAgMmtGdYyT9sBAADciwjJAAAATK5imWJyvsV3hc6Wa+0AAADuV4RkAAAA0JF/Pp5jUOZsubYfAADgfsbE/QAAAJB0LQiL+/Oy2s/YoL+u2uThYtGaYS3pQQYAAEyBkAwAAAAOFcsUU8zEDoVdBgAAQIFjuCUAAAAAAABMj5AMAAAAAAAApkdIBgAAAAAAANMjJAMAAAAAAIDpEZIBAAAAAADA9AjJAAAAAAAAYHqEZAAAAAAAADA9QjIAAAAAAACYHiEZAAAAAAAATI+QDAAAAAAAAKZHSAYAAAAAAADTIyQDAAAAAACA6RGSAQAAAAAAwPQIyQAAAAAAAGB6hGQAAAAAAAAwPUIyAAAAAAAAmB4hGQAAAAAAAEyPkAwAAAAAAACmR0gGAAAAAAAA0yMkAwAAAAAAgOkRkgEAAAAAAMD0CMkAAAAAAABgeoRkAAAAAAAAMD1CMgAAAAAAAJgeIRkAAAAAAABMj5AMAAAAAAAApkdIBgAAAAAAANMjJAMAAAAAAIDpEZIBAAAAAADA9AjJAAAAAAAAYHqEZAAAAAAAADA9QjIAAAAAAACYHiEZAAAoMIZhyDCMwi4DAAAAyMK5sAsAAADm8cEHH8jHx6ewywAAAACyICQDAAAFZvDgwYVdAgAAAJAthlsCAIACc/1wS8Mw5OTk5NhmGIbc3NwKuUIAAACYFSEZAAAoNDabTS4uLnr66adlsViUmpqqJ598srDLAgAAgAkx3BIAAOSbs8kpeuqDzUq4dFXexV2ybZOamipJSklJkbu7u7799tuCLBEAAACQREgGAADySd1x3yn5Sprj/eXE9CxtLJb/dWrPGGqZnp61HQAAAJDfGG4JAADy3I0BWXb7JTnmJ7ue3W7Pt7oAAACAnBCSAQCAPHU2OeWmAZmkW+4HAAAAChohGQAAyFNPfbC5sEsAAAAAbhshGQAAyFMJl64WdgkAAADAbSMkAwAAeSqnVSwlyX/0SvmPXilJenjSOqWlZR52abfbZbPZ8rU+AAAAIDuEZAAAIE8tG9I8T9sBAAAABYGQDAAA5KmyVjdZ3Z1v2sbq7qyyVrcCqggAAAC4NUIyAACQ5/aNa5djUGZ1d9a+ce0KuCIAAADg5m7+a14AAIA7tG9cO51NTtFTH2xWwqWr8i7uomVDmtODDAAAAEUSIRkAAMg3Za1u2vxq68IuAwAAALglhlsCAAAAAADA9AjJAAAAAAAAYHqEZAAAAAAAADA9QjIAAAAAAACYHiEZAAAAAAAATC/fQrK33npLzZo1U7FixeTl5ZWrY+x2u8aOHStfX195eHioTZs2Onz4cH6VCAAAAAAAAEjKx5AsNTVV3bt31+DBg3N9zJQpUzRz5kzNnj1b27ZtU/HixdWuXTtduXIlv8oEAAAAAAAAZNjtdnt+XiAyMlLh4eFKTEy8aTu73S4/Pz+NHDlSo0aNkiQlJSWpfPnyioyMVM+ePXN1veTkZHl6eiopKUlWq/VuywcAAAAAAMA9LLdZUZGZkyw2NlanTp1SmzZtHNs8PT0VHBys6OjoHI9LSUlRcnJyphcAAAAAAABwO4pMSHbq1ClJUvny5TNtL1++vGNfdiZNmiRPT0/Hq0KFCvlaJwAAAAAAAO4/txWSvfrqqzIM46avX3/9Nb9qzdaYMWOUlJTkeP3+++8Fen0AAAAAAADc+5xvp/HIkSPVr1+/m7apVKnSHRXi4+MjSTp9+rR8fX0d20+fPq369evneJybm5vc3Nzu6JoAAAAAAACAdJshWdmyZVW2bNl8KSQwMFA+Pj6KiopyhGLJycnatm3bba2QCQAAAAAAANyufJuTLC4uTnv37lVcXJzS09O1d+9e7d27VxcvXnS0qVGjhpYtWyZJMgxD4eHhevPNN/XNN99o//796tOnj/z8/NS5c+f8KhMAAAAAAAC4vZ5kt2Ps2LGaN2+e432DBg0kST/++KNCQ0MlSYcOHVJSUpKjzSuvvKJLly5p0KBBSkxMVPPmzbVmzRq5u7vnV5kAAAAAAACADLvdbi/sIvJScnKyPD09lZSUJKvVWtjlAAAAAAAAoBDlNivKt+GWAAAAAAAAwL2CkAwAAAAAAACmR0gGAAAAAAAA0yMkAwAAAAAAgOkRkgEAAAAAAMD0CMkAAAAAAABgeoRkAAAAAAAAMD1CMgAAAAAAAJgeIRkAAAAAAABMj5AMAAAAAAAApkdIBgAAAAAAANMjJAMAAACAPNSpUydZLBYZhiHDMOTi4qLY2FhZrVa5u7srMDDQsa9UqVJKSEhwHNu3b185Ozs79ru5uWnhwoWZzr9y5UoVL17c0cbJyUmvv/66Y3+3bt0yXT8wMFAXL14ssPsHgHsVIRkAAAAA5JG1a9dq5cqVqlWrlpYuXaqpU6eqbt26stlskqSUlBSdPn1aM2fOVK9evZSYmKi6des6jk9MTNQTTzyhefPmafLkyZKkPn36KDU1VZJ09OhRderUSSkpKRo+fLjmzZun9u3bKy0tTZIUHh6uL7/8Ui1atNCnn36q559/XseOHVOdOnUK+JMAgHuPYbfb7YVdRF5KTk6Wp6enkpKSZLVaC7scAAAAACbwV2q6/rk6Rt/Mm6V9yz/Sos+WqlfP7pnaWK1WXbhwQcePH1fFihUlSY0bN9bOnTuVkpIiV1fXLOf96aefFBISomnTpik8PFzBwcHavn27du7cqYceeihLexcXFwUEBOjw4cOOba1bt9YPP/yg++xHPwDItdxmRfQkAwAAAIC7EDZ/h2qOXaMFP8UpIbCtZHHSM72elrOrm0JCQrR3715HW2dnZ0dAJkndu18L0lasWCFJioyMzDSUMiQkRJK0Y8cOSdLhw4fl5OSUbUAmSWlpaTpy5IjjeMMw9MMPP0iS4uLi8vzeAeB+4lzYBQAAAADAvSps/g6tjTnjeO/k6qEHR3ylpA2RuvzrZv300zY1aNBAixcvztX5+vfvLxcXF4WFhalBgwZKS0vTyy+/rCtXrki61lPsVqpXr67XXnsty3YfH59c3hUAmBMhGQAAAADcgb9S0zMFZBmcnJzk3WqAvFsNUHrqX/pjWndNnDhR0rWeXidOnNADDzwgSfriiy8kXZvsf/v27ZKkIUOGaPr06ZKkESNGZDp3jRo1tHHjRu3atSvb3mQWi0Xx8fHq06dPnt0nAJgFIRkAAAAA3IF/ro7Jsu38hvm6/OsmFQ8KlUv5QF38ea0kqUGDBo7hjtWqVdPbb7+t6Oho7dixQ35+fnJ1dVX9+vUlSR9//LEaNmyorVu3as6cOZnOv2zZMpUuXVrBwcEKDw9Xw4YNtWTJEtWoUUNvv/22nnvuOc2bN0+BgYEaMWKEnJ2d9fXXX+vAgQP6/fff8/cDAYB7HCEZAAAAANyBY+cuZ9nmVLK00pLPKGnrZ45tpSpU04IFC2S1WuXm5qYyZcropZdekiR5eno65ixzdXXVwIED9fHHH6tv374yDEN/+9vfNGvWLMe5vL299cUXX+jZZ5/V1KlTJV3rPZaxQmZkZKQkacGCBXr55Zcd+5s0aZLn9w8A9xtWtwQAAACAOxCxfL8W/HTryfCfa1pREzvXkdVqVWpqqmN+MQBAwWB1SwAAAADIR691DMrTdgCAwkVIBgAAAAB3wMPVSW2Dyt20TdugcvJwdSqgigAAd4OQDAAAAADu0Jw+jXMMytoGldOcPo0d75OTkxlqCQBFGBP3AwAAAMBdmNOnsf5KTdc/V8fo2LnLCihdTK91DKIHGQDcYwjJAAAAAOAuebg6aWLnOoVdBgDgLjDcEgAAAAAAAKZHSAYAAAAAAADTIyQDAAAAAACA6RGSAQAAAAAAwPQIyQAAAAAAAGB6hGQAAAAAAAAwPUIyAAAAAAAAmB4hGQAAAAAAAEyPkAwAAAAAAACmR0gGAAAAAAAA0yMkAwAAAAAAgOkRkgEAAAAAAMD0CMkAAAAAAABgeoRkAAAAAAAAMD1CMgAAAAAAAJgeIRkAAAAAAABMj5AMAAAAAAAApkdIBgAAAAAAANMjJAMAAAAAAIDpEZIBAAAAAADA9AjJAAAAAAAAYHqEZAAAAAAAADA9QjIAAAAAAACYHiEZAAAAAAAATI+QDAAAAAAAAKZHSAYAAAAAAADTIyQDAAAAAACA6RGSAQAAAAAAwPQIyQAAAAAAAGB6hGQAAAAAAAAwPUIyAAAAAAAAmB4hGQAAAAAAAEyPkAwAAAAAAACmR0gGAAAAAAAA0yMkAwAAAAAAgOkRkgEAAAAAAMD0CMkAAAAAAABgeoRkAAAAAAAAMD1CMgAAAAAAAJgeIRkAAAAAAABMj5AMAAAAAAAApkdIBgAAAAAAANMjJAMAAAAAAIDpEZIBAAAAAADA9JwLu4C8ZrfbJUnJycmFXAkAAAAAAAAKW0ZGlJEZ5eS+C8kuXLggSapQoUIhVwIAAAAAAICi4sKFC/L09Mxxv2G/VYx2j7HZbDp58qRKliwpwzAKuxzTSE5OVoUKFfT777/LarUWdjnAPYnnCMgbPEvA3eM5Au4ezxFw9/LqObLb7bpw4YL8/PxkseQ889h915PMYrHowQcfLOwyTMtqtfIFALhLPEdA3uBZAu4ezxFw93iOgLuXF8/RzXqQZWDifgAAAAAAAJgeIRkAAAAAAABMj5AMecLNzU1vvPGG3NzcCrsU4J7FcwTkDZ4l4O7xHAF3j+cIuHsF/RzddxP3AwAAAAAAALeLnmQAAAAAAAAwPUIyAAAAAAAAmB4hGQAAAAAAAEyPkAwAAAAAAACmR0gGAAAAAAAA0yMkwx1766231KxZMxUrVkxeXl65OsZut2vs2LHy9fWVh4eH2rRpo8OHD+dvoUARlpCQoN69e8tqtcrLy0sDBgzQxYsXb3pMaGioDMPI9HrxxRcLqGKg8P3rX/9SQECA3N3dFRwcrO3bt9+0/eeff64aNWrI3d1dderU0erVqwuoUqBou51nKTIyMsvXHnd39wKsFihaNm7cqE6dOsnPz0+GYWj58uW3PGb9+vVq2LCh3NzcVKVKFUVGRuZ7nUBRdrvP0fr167N8LTIMQ6dOncqzmgjJcMdSU1PVvXt3DR48ONfHTJkyRTNnztTs2bO1bds2FS9eXO3atdOVK1fysVKg6Ordu7cOHDigtWvXauXKldq4caMGDRp0y+PCwsIUHx/veE2ZMqUAqgUK35IlSzRixAi98cYb2r17t+rVq6d27drpzJkz2bbfunWrevXqpQEDBmjPnj3q3LmzOnfurF9++aWAKweKltt9liTJarVm+tpz/PjxAqwYKFouXbqkevXq6V//+leu2sfGxurxxx/Xo48+qr179yo8PFwDBw7Ud999l8+VAkXX7T5HGQ4dOpTp61G5cuXyrCbDbrfb8+xsMKXIyEiFh4crMTHxpu3sdrv8/Pw0cuRIjRo1SpKUlJSk8uXLKzIyUj179iyAaoGi4+DBgwoKCtKOHTvUqFEjSdKaNWvUsWNH/fHHH/Lz88v2uNDQUNWvX1/Tp08vwGqBoiE4OFiNGzfWrFmzJEk2m00VKlTQSy+9pFdffTVL+x49eujSpUtauXKlY1vTpk1Vv359zZ49u8DqBoqa232Wcvv9HmBGhmFo2bJl6ty5c45tRo8erVWrVmX6JU3Pnj2VmJioNWvWFECVQNGWm+do/fr1evTRR3X+/Plcj2a7XfQkQ4GJjY3VqVOn1KZNG8c2T09PBQcHKzo6uhArAwpHdHS0vLy8HAGZJLVp00YWi0Xbtm276bGffvqpypQpo9q1a2vMmDG6fPlyfpcLFLrU1FTt2rUr09cRi8WiNm3a5Ph1JDo6OlN7SWrXrh1fd2Bqd/IsSdLFixfl7++vChUq6Mknn9SBAwcKolzgvsDXIyDv1K9fX76+vmrbtq22bNmSp+d2ztOzATeRMU64fPnymbaXL18+T8cQA/eKU6dOZeka7OzsLG9v75s+E88884z8/f3l5+enffv2afTo0Tp06JC++uqr/C4ZKFR//vmn0tPTs/068uuvv2Z7zKlTp/i6A9zgTp6l6tWr6z//+Y/q1q2rpKQkvfvuu2rWrJkOHDigBx98sCDKBu5pOX09Sk5O1l9//SUPD49Cqgy4d/j6+mr27Nlq1KiRUlJS9PHHHys0NFTbtm1Tw4YN8+QahGTI5NVXX9Xbb7990zYHDx5UjRo1Cqgi4N6T2+foTl0/Z1mdOnXk6+ur1q1b6+jRo6pcufIdnxcAgJyEhIQoJCTE8b5Zs2aqWbOm/v3vf2vixImFWBkAwCyqV6+u6tWrO943a9ZMR48e1bRp07RgwYI8uQYhGTIZOXKk+vXrd9M2lSpVuqNz+/j4SJJOnz4tX19fx/bTp0+rfv36d3ROoCjK7XPk4+OTZYLktLQ0JSQkOJ6X3AgODpYkHTlyhJAM97UyZcrIyclJp0+fzrT99OnTOT4zPj4+t9UeMIM7eZZu5OLiogYNGujIkSP5USJw38np65HVaqUXGXAXmjRpos2bN+fZ+QjJkEnZsmVVtmzZfDl3YGCgfHx8FBUV5QjFkpOTtW3btttaIRMo6nL7HIWEhCgxMVG7du3SQw89JEn64YcfZLPZHMFXbuzdu1eSMoXPwP3I1dVVDz30kKKiohyTutpsNkVFRWno0KHZHhMSEqKoqCiFh4c7tq1duzZTjxjAbO7kWbpRenq69u/fr44dO+ZjpcD9IyQkRKtXr860ja9HwN3bu3dvnv4cREiGOxYXF6eEhATFxcUpPT3d8YN6lSpVVKJECUlSjRo1NGnSJD311FMyDEPh4eF68803VbVqVQUGBioiIkJ+fn43XcECuF/VrFlT7du3V1hYmGbPnq2rV69q6NCh6tmzp2NlyxMnTqh169aaP3++mjRpoqNHj2rRokXq2LGjSpcurX379mn48OF65JFHVLdu3UK+IyD/jRgxQn379lWjRo3UpEkTTZ8+XZcuXVL//v0lSX369NEDDzygSZMmSZKGDRumli1baurUqXr88ce1ePFi7dy5Ux999FFh3gZQ6G73WZowYYKaNm2qKlWqKDExUe+8846OHz+ugQMHFuZtAIXm4sWLmXpSxsbGau/evfL29lbFihU1ZswYnThxQvPnz5ckvfjii5o1a5ZeeeUVPf/88/rhhx+0dOlSrVq1qrBuASh0t/scTZ8+XYGBgapVq5auXLmijz/+WD/88IO+//77vCvKDtyhvn372iVlef3444+ONpLsc+fOdby32Wz2iIgIe/ny5e1ubm721q1b2w8dOlTwxQNFxLlz5+y9evWylyhRwm61Wu39+/e3X7hwwbE/NjY203MVFxdnf+SRR+ze3t52Nzc3e5UqVex///vf7UlJSYV0B0DBe//99+0VK1a0u7q62ps0aWL/6aefHPtatmxp79u3b6b2S5cutVerVs3u6upqr1Wrln3VqlUFXDFQNN3OsxQeHu5oW758eXvHjh3tu3fvLoSqgaLhxx9/zPZnoYznpm/fvvaWLVtmOaZ+/fp2V1dXe6VKlTL9nASY0e0+R2+//ba9cuXKdnd3d7u3t7c9NDTU/sMPP+RpTYbdbrfnXeQGAAAAAAAA3HsshV0AAAAAAAAAUNgIyQAAAAAAAGB6hGQAAAAAAAAwPUIyAAAAAAAAmB4hGQAAAAAAAEyPkAwAAAAAAACmR0gGAAAAAAAA0yMkAwAAAAAAgOkRkgEAAAAAAMD0CMkAAAAAAABgeoRkAAAAAAAAML3/B7EyxAqhL8d6AAAAAElFTkSuQmCC\n"
          },
          "metadata": {}
        }
      ],
      "source": [
        "if __name__ == '__main__':\n",
        "  import matplotlib.pyplot as plt  # for plotting graphs\n",
        "  def plot_semantic_space(dim1, dim2, words, weights, word_to_index, num_vocabs):\n",
        "    \"\"\"\n",
        "    Plots dim1 against dim2 of the embedding of the words.\n",
        "    Args:\n",
        "      dim1: the first dimension to slice\n",
        "      dim2: the second dimension to slice\n",
        "      words: a list of words to plot\n",
        "      weights: the embeddings learned by a model\n",
        "    \"\"\"\n",
        "    coordinates = get_embeddings(words, weights, word_to_index, num_vocabs)[:, [dim1,dim2]]\n",
        "    x_points = coordinates[:, 0]\n",
        "    y_points = coordinates[:, 1]\n",
        "\n",
        "    # use matplotlib to plot a scatter plot of the points\n",
        "    plt.figure(figsize=(15, 15))\n",
        "    plt.scatter(x_points, y_points)\n",
        "    # label the points with the word corresponding to it\n",
        "    for i in range(0, len(words)):\n",
        "      plt.annotate(text=words[i], xy=coordinates[i])\n",
        "\n",
        "  # extract the weights from the model\n",
        "  weights_cbow = model_cbow.layers[0].get_weights()[0]\n",
        "  weights_skipgram = model_skipgram.layers[0].get_weights()[0]\n",
        "  # visualization\n",
        "  plot_semantic_space(10, 20, words_in_order, weights_cbow, word_to_index, num_vocabs)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kKwWwNngh8WI"
      },
      "source": [
        "We can also use the skipgram method to find the closest context words!  Let's try to see what context words are closest to the words \"open\", \"set\", and \"space\"."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Y3Yj2bCxupUf",
        "outputId": "3145ef28-ac89-4543-9657-8c9941640c4f"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "The 10 closest context words for 'open' are: \t ['an', 'quiz', 'except', 'understand', 'swap', 'we', '001', 'is', 'source', '01']\n",
            "The 10 closest context words for 'set' are: \t ['full', 'theory', 'a', 'the', 'options', 'zeroes', 'term', 'isolated', '01', 'each']\n",
            "The 10 closest context words for 'space' are: \t ['3-dimensional', 'projected', 'is', 'position', 'riemann', 'message', 'under', 'anything', 'means', 'function']\n"
          ]
        }
      ],
      "source": [
        "def context_words(word, top=10):\n",
        "  \"\"\"\n",
        "  Finds the top few (default is 10) context words closest to the input word\n",
        "  Args:\n",
        "    word: the target word\n",
        "    top:  specifies how many context words to find\n",
        "  Returns:\n",
        "    a list strings, which are the top few context words\n",
        "  \"\"\"\n",
        "  # get the one-hot encoding of the word\n",
        "  word_1hot = one_hot(np.array([word_to_index[word]]), num_vocabs)\n",
        "  # get the model output\n",
        "  output = model_cbow(word_1hot)[0]\n",
        "  # get an array for the would-be sorted the output, from closest context words to furthest away context words\n",
        "  sorted = np.argsort(-output)\n",
        "  # check out which words they are\n",
        "  return [index_to_word[s] for s in sorted][0:top]\n",
        "\n",
        "if __name__ == '__main__':\n",
        "  for w in ['open', 'set', 'space']:\n",
        "    print(f\"The 10 closest context words for '{w}' are: \\t {context_words(w)}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eINLnzeSi9YD"
      },
      "source": [
        "Disregarding the words without much meaning (a, an, the, is, of, etc.), we can see that\n",
        "\n",
        "- `\"open\"` is close to `\"sets\"`,\n",
        "- `\"set\"` is close to `\"open\"`, and\n",
        "- `\"space\"` is close to `\"metric\"` and `\"topological\"`.  \n",
        "\n",
        "Looking back at the original text, this makes sense because, for example, `\"open sets\"` is a commonly appearing phrase, and so are `\"metric space\"` and `\"topological space\"`."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8kLjOWNNS_J7"
      },
      "source": [
        "# **Section 5**  Training on the Wikipedia dataset (no tasks to complete)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TrPYjIsmomWD"
      },
      "source": [
        "To speed up training and make the process easier to debug, we used a sample text paragraph on open sets to train our Word2Vec model.  In practice, training is done using huge amounts of text, such as all Wikipedia articles instead.  There are two advantages to doing so:\n",
        "\n",
        "- The model learns a much larger number of vocabs\n",
        "- The embeddings are much more accurate because of the richer number of context words we can use during training.\n",
        "\n",
        "Due to Google Colab's usage limits, it is only possible to train around 50 articles before exceeding Colab's usage limits.  Here, we provide some code for you to try this out.  It takes around 10 minutes to train 50 articles for 100 epochs using Colab's GPUs."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "S3Lu49s9qlok",
        "outputId": "c4b30b88-6720-42f7-c18e-a108289eaa4c"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: datasets in /usr/local/lib/python3.10/dist-packages (2.15.0)\n",
            "Requirement already satisfied: numpy>=1.17 in /usr/local/lib/python3.10/dist-packages (from datasets) (1.23.5)\n",
            "Requirement already satisfied: pyarrow>=8.0.0 in /usr/local/lib/python3.10/dist-packages (from datasets) (9.0.0)\n",
            "Requirement already satisfied: pyarrow-hotfix in /usr/local/lib/python3.10/dist-packages (from datasets) (0.6)\n",
            "Requirement already satisfied: dill<0.3.8,>=0.3.0 in /usr/local/lib/python3.10/dist-packages (from datasets) (0.3.1.1)\n",
            "Requirement already satisfied: pandas in /usr/local/lib/python3.10/dist-packages (from datasets) (1.5.3)\n",
            "Requirement already satisfied: requests>=2.19.0 in /usr/local/lib/python3.10/dist-packages (from datasets) (2.31.0)\n",
            "Requirement already satisfied: tqdm>=4.62.1 in /usr/local/lib/python3.10/dist-packages (from datasets) (4.66.1)\n",
            "Requirement already satisfied: xxhash in /usr/local/lib/python3.10/dist-packages (from datasets) (3.4.1)\n",
            "Requirement already satisfied: multiprocess in /usr/local/lib/python3.10/dist-packages (from datasets) (0.70.15)\n",
            "Requirement already satisfied: fsspec[http]<=2023.10.0,>=2023.1.0 in /usr/local/lib/python3.10/dist-packages (from datasets) (2023.6.0)\n",
            "Requirement already satisfied: aiohttp in /usr/local/lib/python3.10/dist-packages (from datasets) (3.8.6)\n",
            "Requirement already satisfied: huggingface-hub>=0.18.0 in /usr/local/lib/python3.10/dist-packages (from datasets) (0.19.4)\n",
            "Requirement already satisfied: packaging in /usr/local/lib/python3.10/dist-packages (from datasets) (23.2)\n",
            "Requirement already satisfied: pyyaml>=5.1 in /usr/local/lib/python3.10/dist-packages (from datasets) (6.0.1)\n",
            "Requirement already satisfied: attrs>=17.3.0 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets) (23.1.0)\n",
            "Requirement already satisfied: charset-normalizer<4.0,>=2.0 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets) (3.3.2)\n",
            "Requirement already satisfied: multidict<7.0,>=4.5 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets) (6.0.4)\n",
            "Requirement already satisfied: async-timeout<5.0,>=4.0.0a3 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets) (4.0.3)\n",
            "Requirement already satisfied: yarl<2.0,>=1.0 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets) (1.9.2)\n",
            "Requirement already satisfied: frozenlist>=1.1.1 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets) (1.4.0)\n",
            "Requirement already satisfied: aiosignal>=1.1.2 in /usr/local/lib/python3.10/dist-packages (from aiohttp->datasets) (1.3.1)\n",
            "Requirement already satisfied: filelock in /usr/local/lib/python3.10/dist-packages (from huggingface-hub>=0.18.0->datasets) (3.13.1)\n",
            "Requirement already satisfied: typing-extensions>=3.7.4.3 in /usr/local/lib/python3.10/dist-packages (from huggingface-hub>=0.18.0->datasets) (4.5.0)\n",
            "Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests>=2.19.0->datasets) (3.4)\n",
            "Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.10/dist-packages (from requests>=2.19.0->datasets) (2.0.7)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests>=2.19.0->datasets) (2023.7.22)\n",
            "Collecting dill<0.3.8,>=0.3.0 (from datasets)\n",
            "  Using cached dill-0.3.7-py3-none-any.whl (115 kB)\n",
            "Requirement already satisfied: python-dateutil>=2.8.1 in /usr/local/lib/python3.10/dist-packages (from pandas->datasets) (2.8.2)\n",
            "Requirement already satisfied: pytz>=2020.1 in /usr/local/lib/python3.10/dist-packages (from pandas->datasets) (2023.3.post1)\n",
            "Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.10/dist-packages (from python-dateutil>=2.8.1->pandas->datasets) (1.16.0)\n",
            "Installing collected packages: dill\n",
            "  Attempting uninstall: dill\n",
            "    Found existing installation: dill 0.3.1.1\n",
            "    Uninstalling dill-0.3.1.1:\n",
            "      Successfully uninstalled dill-0.3.1.1\n",
            "\u001b[31mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n",
            "apache-beam 2.52.0 requires dill<0.3.2,>=0.3.1.1, but you have dill 0.3.7 which is incompatible.\u001b[0m\u001b[31m\n",
            "\u001b[0mSuccessfully installed dill-0.3.7\n",
            "Requirement already satisfied: apache_beam in /usr/local/lib/python3.10/dist-packages (2.52.0)\n",
            "Requirement already satisfied: mwparserfromhell in /usr/local/lib/python3.10/dist-packages (0.6.5)\n",
            "Requirement already satisfied: crcmod<2.0,>=1.7 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (1.7)\n",
            "Requirement already satisfied: orjson<4,>=3.9.7 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (3.9.10)\n",
            "Collecting dill<0.3.2,>=0.3.1.1 (from apache_beam)\n",
            "  Using cached dill-0.3.1.1-py3-none-any.whl\n",
            "Requirement already satisfied: cloudpickle~=2.2.1 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (2.2.1)\n",
            "Requirement already satisfied: fastavro<2,>=0.23.6 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (1.9.0)\n",
            "Requirement already satisfied: fasteners<1.0,>=0.3 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (0.19)\n",
            "Requirement already satisfied: grpcio!=1.48.0,<2,>=1.33.1 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (1.59.2)\n",
            "Requirement already satisfied: hdfs<3.0.0,>=2.1.0 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (2.7.3)\n",
            "Requirement already satisfied: httplib2<0.23.0,>=0.8 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (0.22.0)\n",
            "Requirement already satisfied: js2py<1,>=0.74 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (0.74)\n",
            "Requirement already satisfied: jsonschema<5.0.0,>=4.0.0 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (4.19.2)\n",
            "Requirement already satisfied: numpy<1.25.0,>=1.14.3 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (1.23.5)\n",
            "Requirement already satisfied: objsize<0.7.0,>=0.6.1 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (0.6.1)\n",
            "Requirement already satisfied: packaging>=22.0 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (23.2)\n",
            "Requirement already satisfied: pymongo<5.0.0,>=3.8.0 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (4.6.0)\n",
            "Requirement already satisfied: proto-plus<2,>=1.7.1 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (1.22.3)\n",
            "Requirement already satisfied: protobuf!=4.0.*,!=4.21.*,!=4.22.0,!=4.23.*,!=4.24.*,<4.26.0,>=3.20.3 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (3.20.3)\n",
            "Requirement already satisfied: pydot<2,>=1.2.0 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (1.4.2)\n",
            "Requirement already satisfied: python-dateutil<3,>=2.8.0 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (2.8.2)\n",
            "Requirement already satisfied: pytz>=2018.3 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (2023.3.post1)\n",
            "Requirement already satisfied: regex>=2020.6.8 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (2023.6.3)\n",
            "Requirement already satisfied: requests<3.0.0,>=2.24.0 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (2.31.0)\n",
            "Requirement already satisfied: typing-extensions>=3.7.0 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (4.5.0)\n",
            "Requirement already satisfied: zstandard<1,>=0.18.0 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (0.22.0)\n",
            "Requirement already satisfied: pyarrow<12.0.0,>=3.0.0 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (9.0.0)\n",
            "Requirement already satisfied: pyarrow-hotfix<1 in /usr/local/lib/python3.10/dist-packages (from apache_beam) (0.6)\n",
            "Requirement already satisfied: docopt in /usr/local/lib/python3.10/dist-packages (from hdfs<3.0.0,>=2.1.0->apache_beam) (0.6.2)\n",
            "Requirement already satisfied: six>=1.9.0 in /usr/local/lib/python3.10/dist-packages (from hdfs<3.0.0,>=2.1.0->apache_beam) (1.16.0)\n",
            "Requirement already satisfied: pyparsing!=3.0.0,!=3.0.1,!=3.0.2,!=3.0.3,<4,>=2.4.2 in /usr/local/lib/python3.10/dist-packages (from httplib2<0.23.0,>=0.8->apache_beam) (3.1.1)\n",
            "Requirement already satisfied: tzlocal>=1.2 in /usr/local/lib/python3.10/dist-packages (from js2py<1,>=0.74->apache_beam) (5.2)\n",
            "Requirement already satisfied: pyjsparser>=2.5.1 in /usr/local/lib/python3.10/dist-packages (from js2py<1,>=0.74->apache_beam) (2.7.1)\n",
            "Requirement already satisfied: attrs>=22.2.0 in /usr/local/lib/python3.10/dist-packages (from jsonschema<5.0.0,>=4.0.0->apache_beam) (23.1.0)\n",
            "Requirement already satisfied: jsonschema-specifications>=2023.03.6 in /usr/local/lib/python3.10/dist-packages (from jsonschema<5.0.0,>=4.0.0->apache_beam) (2023.11.1)\n",
            "Requirement already satisfied: referencing>=0.28.4 in /usr/local/lib/python3.10/dist-packages (from jsonschema<5.0.0,>=4.0.0->apache_beam) (0.31.0)\n",
            "Requirement already satisfied: rpds-py>=0.7.1 in /usr/local/lib/python3.10/dist-packages (from jsonschema<5.0.0,>=4.0.0->apache_beam) (0.13.0)\n",
            "Requirement already satisfied: dnspython<3.0.0,>=1.16.0 in /usr/local/lib/python3.10/dist-packages (from pymongo<5.0.0,>=3.8.0->apache_beam) (2.4.2)\n",
            "Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.10/dist-packages (from requests<3.0.0,>=2.24.0->apache_beam) (3.3.2)\n",
            "Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.10/dist-packages (from requests<3.0.0,>=2.24.0->apache_beam) (3.4)\n",
            "Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.10/dist-packages (from requests<3.0.0,>=2.24.0->apache_beam) (2.0.7)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.10/dist-packages (from requests<3.0.0,>=2.24.0->apache_beam) (2023.7.22)\n",
            "Installing collected packages: dill\n",
            "  Attempting uninstall: dill\n",
            "    Found existing installation: dill 0.3.7\n",
            "    Uninstalling dill-0.3.7:\n",
            "      Successfully uninstalled dill-0.3.7\n",
            "\u001b[31mERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.\n",
            "multiprocess 0.70.15 requires dill>=0.3.7, but you have dill 0.3.1.1 which is incompatible.\u001b[0m\u001b[31m\n",
            "\u001b[0mSuccessfully installed dill-0.3.1.1\n"
          ]
        }
      ],
      "source": [
        "if __name__ == '__main__':\n",
        "  # uncommment and run to install necessary libs\n",
        "  !pip install datasets # for loading wikipedia dataset\n",
        "  !pip install apache_beam mwparserfromhell # for loading wikipedia dataset\n",
        "  from datasets import load_dataset\n",
        "  data = load_dataset(\"wikipedia\", \"20220301.simple\")\n",
        "  data = data['train']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Ad2fulZUrOAi"
      },
      "outputs": [],
      "source": [
        "# data is a list of 205328 dictionaries.\n",
        "# data[index] is a dictionary containing keys ['id', 'url', 'title', 'text']\n",
        "# here are all the article titles that are available\n",
        "\n",
        "if __name__ == '__main__':\n",
        "  titles = [d['title'].lower() for d in data]\n",
        "  titles"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FbWLLZYA-KXb",
        "outputId": "6b361a63-2863-4e3c-904d-336408a25a5f"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "4744 \t [function (mathematics)] \n",
            "\t\t In mathematics, a function is a mathematical object that produces an output, when given an input (wh...\n",
            "8685 \t [hash function] \n",
            "\t\t The Hash function is a function. When a computer program is written, very often, large amounts of da...\n",
            "13586 \t [euler's totient function] \n",
            "\t\t In number theory, the totient of a positive integer n is the number of positive integers smaller tha...\n",
            "14001 \t [inverse function] \n",
            "\t\t An inverse function is a concept of mathematics. A function will calculate some output , given some ...\n",
            "14003 \t [computable function] \n",
            "\t\t A computable function is an idea from computer science. A function is computable if there is an algo...\n",
            "14346 \t [function] \n",
            "\t\t Function may mean:  professional or official position: Official  Function (mathematics), the relatio...\n",
            "14580 \t [trigonometric function] \n",
            "\t\t In mathematics, the trigonometric functions are a set of functions which relate angles to the sides ...\n",
            "16294 \t [prime counting function] \n",
            "\t\t In mathematics, the prime counting function is the function counting the number of prime numbers les...\n",
            "25619 \t [exponential function] \n",
            "\t\t In mathematics, the exponential function is a function that grows quicker and quicker. More precisel...\n",
            "30722 \t [limit of a function] \n",
            "\t\t In calculus, a branch of mathematics, the limit of a function is the behavior of a certain function ...\n",
            "34985 \t [cryptographic hash function] \n",
            "\t\t A cryptographic hash function is a hash function which takes an input (or 'message') and returns a f...\n",
            "37275 \t [functional illiteracy] \n",
            "\t\t Functional illiteracy means that a person is not able to read, write, and do math well in regular li...\n",
            "47604 \t [dirac delta function] \n",
            "\t\t The Dirac delta function, often written as , is a made-up concept by mathematician Paul Dirac. It is...\n",
            "48341 \t [one-way function] \n",
            "\t\t In cryptography, a one-way function is a function that is very simple to compute (figure out) in one...\n",
            "48342 \t [sha hash functions] \n",
            "\t\t In cryptography, the Secure Hashing Algorithms are a group of cryptographic hash functions released ...\n",
            "48372 \t [riemann zeta function] \n",
            "\t\t In mathematics, the Riemann zeta function is an important function in number theory. It is related t...\n",
            "54734 \t [heaviside function] \n",
            "\t\t The Heaviside function, often written as H(x), is a non-continuous function whose value is zero for ...\n",
            "57684 \t [continuous function] \n",
            "\t\t A mathematical function is called continuous if, roughly said, a small change in the input only caus...\n",
            "59006 \t [anonymous function] \n",
            "\t\t In computer science and mathematics, an anonymous function is a function with no name.  Usually, a f...\n",
            "59976 \t [periodic function] \n",
            "\t\t In mathematics, a periodic function is a function which has a repeated pattern (a period) which is c...\n",
            "62226 \t [functional group] \n",
            "\t\t In organic chemistry, a group of specific atoms, or a specific arrangement of atoms, can form a func...\n",
            "71414 \t [injective function] \n",
            "\t\t In mathematics, a injective function is a function f : A → B with the following property. For every ...\n",
            "71417 \t [surjective function] \n",
            "\t\t In mathematics, a surjective or onto function is a  function f : A → B with the following property. ...\n",
            "72783 \t [functional magnetic resonance imaging] \n",
            "\t\t Functional Magnetic Resonance Imaging (fMRI) is a special type of MRI that makes it possible to watc...\n",
            "73103 \t [functional programming] \n",
            "\t\t Functional programming is a programming paradigm where values are passed around into functions, and ...\n",
            "73355 \t [multivalued function] \n",
            "\t\t A multivalued function (multifunction) is a relation like a function.  A multifunction associates ea...\n",
            "77221 \t [convex function] \n",
            "\t\t In mathematics, a convex function is any function with value of the weighted average of 2 points is ...\n",
            "81473 \t [high-functioning autism] \n",
            "\t\t High-functioning autism (HFA) is an informal term given to autistic people who do not have an intell...\n",
            "82586 \t [wave function] \n",
            "\t\t In quantum mechanics, the Wave function, usually represented by Ψ, or ψ, describes the probability o...\n",
            "86760 \t [function composition] \n",
            "\t\t In mathematics, function composition is a way of making a new function from two other functions thro...\n",
            "88644 \t [gamma function] \n",
            "\t\t In mathematics, the gamma function (Γ(z)) is a key topic in the field of special functions. Γ(z) is ...\n",
            "92842 \t [holomorphic function] \n",
            "\t\t In mathematics a holomorphic function is a function that is defined on the complex numbers, and that...\n",
            "96710 \t [function space] \n",
            "\t\t In mathematics, a function space is a set of functions of a given kind from a set X to a set Y. For ...\n",
            "98679 \t [linear predictor function] \n",
            "\t\t In statistics and in machine learning, a linear predictor function is a linear function (linear comb...\n",
            "98964 \t [loss function] \n",
            "\t\t In mathematics a loss function or cost function is a mathematical function that maps an event, or th...\n",
            "99866 \t [constant function] \n",
            "\t\t In mathematics, a constant function is a function whose output value is the same for every input val...\n",
            "99990 \t [linear function] \n",
            "\t\t In basic mathematics, a linear function is a function whose graph is a straight line in 2-dimensions...\n",
            "100966 \t [erectile disfunction] \n",
            "\t\t Erectile dysfunction (ED, also called impotence) is when a man can't have or keep an erection of the...\n",
            "101043 \t [bijective function] \n",
            "\t\t In mathematics, a bijective function or bijection is a function f : A → B that is both an injection ...\n",
            "101403 \t [network functions virtualization] \n",
            "\t\t Network Functions Virtualization (NFV)   is a computer communication technology that proposes to sub...\n",
            "104993 \t [wavefunction collapse] \n",
            "\t\t When a scientific experiment is done properly, it will give a measurable result.  At each moment, th...\n",
            "109335 \t [lateralization of brain function] \n",
            "\t\t The human brain has two halves, called the left and right hemispheres.  These two halves of the brai...\n",
            "116272 \t [multiple organ dysfunction syndrome] \n",
            "\t\t Multiple organ dysfunction syndrome (MODS), previously known as multiple organ failure (MOF), Total ...\n",
            "117555 \t [probability density function] \n",
            "\t\t In probability and statistics, a probability density function is a function that characterizes any c...\n",
            "117616 \t [root of a function] \n",
            "\t\t In mathematics, a root of a function f is a number x that turns the value of f to 0:  Because polyno...\n",
            "120541 \t [function (biology)] \n",
            "\t\t Function in biology is what a part of the body or cell does in relation to the whole. It is the reas...\n",
            "121797 \t [unary function] \n",
            "\t\t A unary function is a function that takes just a single argument. In contrast, a function that takes...\n",
            "123540 \t [monotonic function] \n",
            "\t\t In algebra, a montonic function is any function whose gradient never changes sign. In simple words, ...\n",
            "127692 \t [euler's homogeneous function theorem] \n",
            "\t\t Euler's theorem is one of the theorems Leonhard Euler stated: There are certain conditions where a f...\n",
            "129563 \t [physical unclonable function] \n",
            "\t\t Physical Unclonable Functions (PUFs) are hardware modules that can generate randomness. Their operat...\n",
            "131975 \t [rational function] \n",
            "\t\t A rational function is a polynomial divided by a polynomial.  is a rational function.  An asymptote ...\n",
            "149253 \t [executive functions] \n",
            "\t\t Higher animals, including humans use their brain to control their behavior. The cognitive processes ...\n",
            "150034 \t [consumption function] \n",
            "\t\t In economics, the consumption function shows a relationship between consumption and disposable incom...\n",
            "150093 \t [key derivation function] \n",
            "\t\t In cryptography, a key derivation function (KDF) makes a long secret key (which is called a \"hash\") ...\n",
            "152955 \t [window function] \n",
            "\t\t In mathematics, a window function is a special function that can be applied to a signal, as it occur...\n",
            "162641 \t [functional analysis] \n",
            "\t\t Functional analysis is a branch of mathematical analysis. This area emerged from the studies of diff...\n",
            "170979 \t [special functions] \n",
            "\t\t Special functions are some mathematical functions used in mathematical analysis or physics. Most of ...\n",
            "171477 \t [matrix function] \n",
            "\t\t In mathematics, a function maps an input value to an output value. In the case of a matrix function,...\n",
            "172056 \t [structural functionalism] \n",
            "\t\t Structural functionalism, or simply functionalism, is a way of making a theory that sees society as ...\n",
            "172060 \t [functionalism] \n",
            "\t\t Functionalism may refer to:   Functionalism (architecture), the idea that architects should design a...\n",
            "174339 \t [sexual dysfunction] \n",
            "\t\t Sexual dysfunction is a medical problem in which a person or a couple have problems during any stage...\n",
            "174654 \t [hyperbolic functions] \n",
            "\t\t \\  In mathematics, hyperbolic functions are analogues of the ordinary trigonometric functions, but d...\n",
            "189434 \t [functionalism (philosophy of mind)] \n",
            "\t\t Functionalism in the philosophy of mind is the doctrine that what makes something a mental state of ...\n",
            "195383 \t [meromorphic function] \n",
            "\t\t A meromorphic function is a function that is holomorphic (defined on the complex numbers, and that c...\n",
            "195440 \t [l-functions and modular forms database] \n",
            "\t\t L-functions and modular forms database is an extensive database of mathematical objects arising in n...\n"
          ]
        }
      ],
      "source": [
        "# You can search for articles you are interested in and extract them for training here\n",
        "if __name__ == '__main__':\n",
        "  keyword = \"function\"  # search for articles with this keyword in the title\n",
        "\n",
        "  import re  # for text manipulation\n",
        "  for i in range(len(data)):\n",
        "    if (keyword.lower() in titles[i]):\n",
        "      title = titles[i]\n",
        "      content = re.sub('\\n', ' ', data[i]['text'][0:100])\n",
        "      print(f\"{i} \\t [{title}] \\n\\t\\t {content}...\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yoz_IYwVNvnu",
        "outputId": "97c4d25b-32b1-4165-c131-241a64121ec1"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "The article titles:\n",
            "['Calculus', 'Differential calculus', 'Function (mathematics)', 'Hash function', \"Euler's totient function\", 'Inverse function', 'Computable function', 'Trigonometric function', 'Lambda calculus', 'Prime counting function', 'Exponential function', 'Limit of a function', 'Cryptographic hash function', 'Dirac delta function', 'One-way function', 'SHA hash functions', 'Riemann zeta function', 'Heaviside Function', 'Continuous function', 'Anonymous function', 'Periodic function', 'Fundamental theorem of calculus', 'Injective function', 'Surjective function', 'Vector calculus', 'Multivariable calculus', 'Functional programming', 'Multivalued function', 'Convex function', 'Wave function', 'Function composition', 'Gamma function', 'Holomorphic function', 'Function space', 'Linear predictor function', 'Loss function', 'Numerical integration', 'Calculus of variations', 'Constant function', 'Linear function', 'Probability density function', 'Root of a function', 'Unary function', 'Monotonic function', \"Euler's homogeneous function theorem\", 'Rational function', 'Open set', 'Key derivation function', 'Window function', 'Functional analysis', 'Special functions', 'Matrix function', 'Hyperbolic functions', 'Meromorphic function', 'Integration by substitution']\n"
          ]
        }
      ],
      "source": [
        "# here are some additional articles related to mathematics.  You can change this to other articles if you want.\n",
        "if __name__ == '__main__':\n",
        "  articles = [    84,   4730,   4744,   8685,  13586,  14001,  14003,  14580,  14947,  16294,\n",
        "              25619,  30722,  34985,  47604,  48341,  48342,  48372,  54734,  57684,  59006,\n",
        "              59976,  63039,  71414,  71417,  71457,  71459,  73103,  73355,  77221,  82586,\n",
        "              86760,  88644,  92842,  96710,  98679,  98964,  99261,  99749,  99866,  99990,\n",
        "              117555, 117616, 121797, 123540, 127692, 131975, 134546, 150093, 152955, 162641,\n",
        "              170979, 171477, 174654, 195383, 199312]\n",
        "  print(\"The article titles:\")\n",
        "  print([data[i]['title'] for i in articles])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "X6CNt3pbSllX"
      },
      "outputs": [],
      "source": [
        "# You can run this cell and comment out the \"text\" variable on open sets in the start of section 2, then run the notebook from the start.\n",
        "# Around 100 epochs is enough for the skipgram model to converge\n",
        "\n",
        "if __name__ == '__main__':\n",
        "  text = ' '.join(data[i]['text'] for i in articles)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HCMLABvjqbC7"
      },
      "source": [
        "# **Appendix** (no tasks to complete)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QFJtei6LkTkv"
      },
      "source": [
        "**2D Semantic Space**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 472
        },
        "id": "Pnz6admh8u3P",
        "outputId": "837ec8f8-6835-4320-af6a-5767a96ab05a"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# 2D Semantic Space\n",
        "# Run this code cell to generate the image\n",
        "\n",
        "if __name__ == '__main__':\n",
        "  import matplotlib.pyplot as plt\n",
        "  import numpy as np\n",
        "\n",
        "  words = {\n",
        "      \"cow\": [3.5, 4],          \"duck\": [3, 3.5],       \"pork\": [5, 4.5],      \"beef\": [5.5, 4],        \"milk\": [4.5, 3],      \"cheese\": [4.75, 3.4],\n",
        "      \"watercress\": [4.5, 0.5], \"spinach\": [5, 0.25],   \"coriander\": [5, 1],   \"llama\": [0.25, 5.25],   \"dog\": [0.5, 5],       \"frog\": [0.75, 4.75]\n",
        "  }\n",
        "\n",
        "  # create a separate list for the words and the coordinates (separation is needed for matplotlib.pyplot)\n",
        "  names, coordinates = [key for key in words.keys()], np.asarray([coord for coord in words.values()])\n",
        "  x_points, y_points = coordinates[:, 0], coordinates[:, 1]\n",
        "\n",
        "  # create a figure and axes object\n",
        "  figure_2d = plt.figure()\n",
        "  axes = figure_2d.add_subplot()\n",
        "\n",
        "  # plot the points on the axes\n",
        "  axes.scatter(x_points, y_points)\n",
        "\n",
        "  # set the x and y axis limits\n",
        "  axes.set_xlim(xmin=0, xmax=6)\n",
        "  axes.set_ylim(ymin=0, ymax=6)\n",
        "\n",
        "  # set the title of the figure\n",
        "  axes.set_title(\"2D Semantic Space\")\n",
        "\n",
        "  # set the title of the x and y axes\n",
        "  axes.set_xlabel(\"food\")\n",
        "  axes.set_ylabel(\"animal\")\n",
        "\n",
        "  # annotate the points\n",
        "  for name, coordinate in zip(names, coordinates):\n",
        "    axes.annotate(text=name, xy=coordinate+0.1)  # text is the label name, xy is the coordinate to place the text\n",
        "\n",
        "  # figure_2d.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5GyJ4KlpkZAz"
      },
      "source": [
        "**3D Semantic Space**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 829
        },
        "id": "sF9wy3GTCdzW",
        "outputId": "f15cf32e-c945-428b-8957-0d06e9cdd640"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1500x1000 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# 3D Semantic Space\n",
        "# Run this code cell to generate the image\n",
        "\n",
        "if __name__ == '__main__':\n",
        "  import matplotlib.pyplot as plt\n",
        "  import numpy as np\n",
        "\n",
        "  words = {\n",
        "      \"cow\": [3.5, 4, 0],          \"duck\": [3, 3.5, 0],       \"pork\": [5, 4.5, 0],      \"beef\": [5.5, 4, 0],        \"milk\": [4.5, 3, 0],      \"cheese\": [4.75, 3.4, 0],\n",
        "      \"watercress\": [4.5, 0.5, 0], \"spinach\": [5, 0.25, 0],   \"coriander\": [5, 1, 0],   \"llama\": [0.25, 5.25, 0],   \"dog\": [0.5, 5, 5],       \"frog\": [0.75, 4.75, 1],\n",
        "      \"cat\": [0, 5, 6],            \"rabbit\": [0, 5.5, 5],     \"hamster\": [0, 5, 4]\n",
        "  }\n",
        "\n",
        "  # create a separate list for the words and the coordinates (separation is needed for matplotlib.pyplot)\n",
        "  names, coordinates = [key for key in words.keys()], np.asarray([coord for coord in words.values()])\n",
        "  x_points, y_points, z_points = coordinates[:, 0], coordinates[:, 1], coordinates[:, 2]\n",
        "\n",
        "  # create a figure and axes object\n",
        "  figure_3d = plt.figure(figsize=(15, 10))  # set figsize to enlarge the plot.  Enlargement is needed because the labels to the data points overlap\n",
        "  axes = figure_3d.add_subplot(projection=\"3d\")\n",
        "\n",
        "  # plot the points on the axes\n",
        "  axes.scatter(x_points, y_points, z_points)\n",
        "\n",
        "  # set the x and y and z axis limits\n",
        "  axes.set_xlim(xmin=0, xmax=6)\n",
        "  axes.set_ylim(ymin=0, ymax=6)\n",
        "  axes.set_zlim(zmin=0, zmax=6)\n",
        "\n",
        "  # set the title of the figure\n",
        "  axes.set_title(\"3D Semantic Space\")\n",
        "\n",
        "  # set the title of the x and y axes\n",
        "  axes.set_xlabel(\"food\")\n",
        "  axes.set_ylabel(\"animal\")\n",
        "  axes.set_zlabel(\"pet\")\n",
        "  axes.zaxis.labelpad=-6  # matplotlib cuts off the z-axis label, adjust this to view z-axis label\n",
        "\n",
        "  # # annotate the points\n",
        "  for name, x, y, z in zip(names, x_points, y_points, z_points):\n",
        "    axes.text(x+0.3, y-0.3, z, name, zdir=(1, 0, 0))  # zdir is a direction vector specifying which direction the text should be pointed at\n",
        "\n",
        "  # figure_3d.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SGZx0oMtZxex"
      },
      "source": [
        "**One-hot encoding**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DU-p9MipkeGd"
      },
      "source": [
        "Let's say that given the properties of some items, we are trying to predict the color of the item.  The output (y) of the training data could be a list of colors as follows:\n",
        "\n",
        "```\n",
        "y = [red, green, yellow, red, red, blue]\n",
        "```\n",
        "\n",
        "Obviously, we need to use numbers to encode the output instead of color names.\n",
        "We could simply map `red` to 0, `green` to 1, `yellow` to 2, and `blue` to 3 for training.  In other words, we use the following vector:\n",
        "\n",
        "```\n",
        "y_encoded = [0, 1, 2, 0, 0, 3]\n",
        "```\n",
        "\n",
        "However, this type of mapping implicitly assigns some form of \"order\" to the colors.  For example, by giving `blue` the label 3 and `yellow` the label 2, the model might somehow pick up some bias that `blue` is better than `yellow`.  To avoid this type of situation, we use one-hot encoding to encode the labels instead.  The one-hot encoding of `y` is a 2D array consisting of only 1s and 0s.  The details of how to create the one-hot encoding of `y` are:\n",
        "\n",
        "- the 0th entry of the 0th row is 1\n",
        "- the 1th entry of the 1st row is 1\n",
        "- the 2th entry of the 2nd row is 1\n",
        "- the 0th entry of the 3rd row is 1\n",
        "- the 0th entry of the 4th row is 1\n",
        "- the 3th entry of the 5th row is 1\n",
        "- all other entries are 0\n",
        "\n",
        "So, the one-hot encoding of `y` is\n",
        "```\n",
        "[[1, 0, 0, 0],\n",
        " [0, 1, 0, 0],\n",
        " [0, 0, 1, 0],\n",
        " [1, 0, 0, 0],\n",
        " [1, 0, 0, 0],\n",
        " [0, 0, 0, 1]]\n",
        "```\n",
        "\n",
        "As another example, if this was the original 1D array,\n",
        "\n",
        "```\n",
        "arr = [0, 2, 5, 1]\n",
        "```\n",
        "\n",
        "and there were 6 classes, then the one-hot encoding of `arr` would be:\n",
        "```\n",
        "[[1, 0, 0, 0, 0, 0],\n",
        " [0, 0, 1, 0, 0, 0],\n",
        " [0, 0, 0, 0, 0, 1],\n",
        " [0, 1, 0, 0, 0, 0]]\n",
        "```\n",
        "\n",
        "Note that if there were some classes that no datum belonged to, then it is possible that some columns contain only 0."
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "toc_visible": true,
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}