Skip_Gram_&_Dimensionality_Reduction_and_Visualization.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.6.9"
    },
    "colab": {
      "name": "Skip Gram & Dimensionality Reduction and Visualization",
      "provenance": [],
      "collapsed_sections": []
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "9a0ec075584699a44c46933457b0a8ba",
          "grade": false,
          "grade_id": "cell-a910b376742d04c0",
          "locked": true,
          "schema_version": 1,
          "solution": false
        },
        "id": "ECD5r2hFKDVd"
      },
      "source": [
        "# Part 1: Skip Gram\n",
        "\n",
        "\n",
        "Available software:\n",
        " - Python's Gensim module: https://radimrehurek.com/gensim/ (install using pip)\n",
        " - Sklearn’s  TSNE module in case you use TSNE to reduce dimension (optional)\n",
        " - Python’s Matplotlib (optional)\n",
        "\n",
        "_Note: The most important hyper parameters of skip-gram/CBOW are vector size and windows size_\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "a09a0bf3042da711c4bf843e9b4a4189",
          "grade": false,
          "grade_id": "cell-bf780e597c0216c8",
          "locked": true,
          "schema_version": 1,
          "solution": false
        },
        "id": "Vsocwry-KDVe",
        "outputId": "d74f3d78-b1b3-4e79-d91a-c0d6926e115e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 479
        }
      },
      "source": [
        "!pip install gensim\n",
        "import os\n",
        "import string\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import re\n",
        "import pprint\n",
        "from urllib.request import urlopen\n",
        "\n",
        "import gensim\n",
        "from gensim import corpora, models, similarities\n",
        "import nltk\n",
        "from nltk import word_tokenize, tokenize\n",
        "nltk.download('punkt')\n",
        "!wget -c \"https://s3.amazonaws.com/dl4j-distribution/GoogleNews-vectors-negative300.bin.gz\""
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: gensim in /usr/local/lib/python3.6/dist-packages (3.6.0)\n",
            "Requirement already satisfied: six>=1.5.0 in /usr/local/lib/python3.6/dist-packages (from gensim) (1.12.0)\n",
            "Requirement already satisfied: smart-open>=1.2.1 in /usr/local/lib/python3.6/dist-packages (from gensim) (1.9.0)\n",
            "Requirement already satisfied: numpy>=1.11.3 in /usr/local/lib/python3.6/dist-packages (from gensim) (1.17.3)\n",
            "Requirement already satisfied: scipy>=0.18.1 in /usr/local/lib/python3.6/dist-packages (from gensim) (1.3.1)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from smart-open>=1.2.1->gensim) (2.21.0)\n",
            "Requirement already satisfied: boto3 in /usr/local/lib/python3.6/dist-packages (from smart-open>=1.2.1->gensim) (1.10.13)\n",
            "Requirement already satisfied: boto>=2.32 in /usr/local/lib/python3.6/dist-packages (from smart-open>=1.2.1->gensim) (2.49.0)\n",
            "Requirement already satisfied: idna<2.9,>=2.5 in /usr/local/lib/python3.6/dist-packages (from requests->smart-open>=1.2.1->gensim) (2.8)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests->smart-open>=1.2.1->gensim) (2019.9.11)\n",
            "Requirement already satisfied: chardet<3.1.0,>=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests->smart-open>=1.2.1->gensim) (3.0.4)\n",
            "Requirement already satisfied: urllib3<1.25,>=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests->smart-open>=1.2.1->gensim) (1.24.3)\n",
            "Requirement already satisfied: jmespath<1.0.0,>=0.7.1 in /usr/local/lib/python3.6/dist-packages (from boto3->smart-open>=1.2.1->gensim) (0.9.4)\n",
            "Requirement already satisfied: botocore<1.14.0,>=1.13.13 in /usr/local/lib/python3.6/dist-packages (from boto3->smart-open>=1.2.1->gensim) (1.13.13)\n",
            "Requirement already satisfied: s3transfer<0.3.0,>=0.2.0 in /usr/local/lib/python3.6/dist-packages (from boto3->smart-open>=1.2.1->gensim) (0.2.1)\n",
            "Requirement already satisfied: docutils<0.16,>=0.10 in /usr/local/lib/python3.6/dist-packages (from botocore<1.14.0,>=1.13.13->boto3->smart-open>=1.2.1->gensim) (0.15.2)\n",
            "Requirement already satisfied: python-dateutil<2.8.1,>=2.1; python_version >= \"2.7\" in /usr/local/lib/python3.6/dist-packages (from botocore<1.14.0,>=1.13.13->boto3->smart-open>=1.2.1->gensim) (2.6.1)\n",
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Package punkt is already up-to-date!\n",
            "--2019-11-12 23:21:11--  https://s3.amazonaws.com/dl4j-distribution/GoogleNews-vectors-negative300.bin.gz\n",
            "Resolving s3.amazonaws.com (s3.amazonaws.com)... 52.216.236.101\n",
            "Connecting to s3.amazonaws.com (s3.amazonaws.com)|52.216.236.101|:443... connected.\n",
            "HTTP request sent, awaiting response... 416 Requested Range Not Satisfiable\n",
            "\n",
            "    The file is already fully retrieved; nothing to do.\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "47031c66b74746d23ccc5e8369446a4b",
          "grade": false,
          "grade_id": "cell-3f89500615a0096f",
          "locked": true,
          "schema_version": 1,
          "solution": false
        },
        "id": "ZF74G9bDKDVh"
      },
      "source": [
        "**Q1 (1 point)** : Find the cosine similarity between the following word pairs\n",
        "\n",
        "- (France, England)\n",
        "- (smaller, bigger)\n",
        "- (England, London)\n",
        "- (France, Rocket)\n",
        "- (big, bigger)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jOXlVGZCwS_w",
        "outputId": "7e94a048-8686-4244-8432-0edf62aa95a4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        }
      },
      "source": [
        "from gensim.models import KeyedVectors\n",
        "model = KeyedVectors.load_word2vec_format('GoogleNews-vectors-negative300.bin.gz', binary=True)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/smart_open/smart_open_lib.py:402: UserWarning: This function is deprecated, use smart_open.open instead. See the migration notes for details: https://github.com/RaRe-Technologies/smart_open/blob/master/README.rst#migrating-to-the-new-open-function\n",
            "  'See the migration notes for details: %s' % _MIGRATION_NOTES_URL\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "nbgrader": {
          "checksum": "4d52dda406c3d8cd5e37d29755f0fb12",
          "grade": false,
          "grade_id": "cell-fbbe575f8f5a6368",
          "locked": false,
          "schema_version": 1,
          "solution": true
        },
        "id": "SZD5ZaMvKDVk",
        "outputId": "74e261ec-2314-4196-b498-acace039f74f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        }
      },
      "source": [
        "#Replace 0 with the code / value; Do not delete this cell\n",
        "similarity_pair1 = model.similarity('France', 'England')\n",
        "similarity_pair2 = model.similarity('smaller', 'bigger')\n",
        "similarity_pair3 = model.similarity('England', 'London')\n",
        "similarity_pair4 = model.similarity('France', 'Rocket')\n",
        "similarity_pair5 = model.similarity('big', 'bigger')\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/gensim/matutils.py:737: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
            "  if np.issubdtype(vec.dtype, np.int):\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "569aa8b664a41d901bf7b0a5e23e9930",
          "grade": true,
          "grade_id": "cell-929d59ed5d67f618",
          "locked": true,
          "points": 1,
          "schema_version": 1,
          "solution": false
        },
        "id": "tFUPLSK7KDVp",
        "outputId": "e90fbb4a-0f12-44f0-ac96-02875723a879",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "#This is an autograded cell, do not edit/delete\n",
        "print(similarity_pair1, similarity_pair2, similarity_pair3, similarity_pair4, similarity_pair5)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.39804944 0.7302272 0.43992856 0.07114174 0.68423855\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "a7f270405ddf9ecbffde36e6c096b818",
          "grade": false,
          "grade_id": "cell-ccd6618b4fac3715",
          "locked": true,
          "schema_version": 1,
          "solution": false
        },
        "id": "ZcqpWCjJKDVs"
      },
      "source": [
        "**Q2 (1 point)** : Write an expression to extract the vector representations of the words: \n",
        "\n",
        "- France\n",
        "- England\n",
        "- smaller\n",
        "- bigger\n",
        "- rocket\n",
        "- big\n",
        "\n",
        "Get only the first 5 elements for each vector representation."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "nbgrader": {
          "checksum": "6b3cecb268eb9440c446cd3de984b7f6",
          "grade": false,
          "grade_id": "cell-00f3d05abb28aa23",
          "locked": false,
          "schema_version": 1,
          "solution": true
        },
        "id": "6pzKlLyjKDVt"
      },
      "source": [
        "#Replace 0 with the code / value to get the first 5 elements of each vector; Do not delete this cell\n",
        "vector_1 = model['France'][:5]\n",
        "vector_2 = model['England'][:5]\n",
        "vector_3 = model['smaller'][:5]\n",
        "vector_4 = model['bigger'][:5]\n",
        "vector_5 = model['rocket'][:5]\n",
        "vector_6 = model['big'][:5]\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "401940f859774b3c1ec48338fa15682e",
          "grade": true,
          "grade_id": "cell-6f34229370fa873f",
          "locked": true,
          "points": 1,
          "schema_version": 1,
          "solution": false
        },
        "id": "Hkj2ROGTKDVv",
        "outputId": "c99bba85-8f7a-4587-b7f2-f6e920de6717",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "source": [
        "#This is an autograded cell, do not edit/delete\n",
        "print(vector_1)\n",
        "print(vector_2)\n",
        "print(vector_3)\n",
        "print(vector_4)\n",
        "print(vector_5)\n",
        "print(vector_6)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[0.04858398 0.07861328 0.32421875 0.03491211 0.07714844]\n",
            "[-0.19824219  0.11523438  0.0625     -0.05834961  0.2265625 ]\n",
            "[-0.05004883  0.03417969 -0.0703125   0.17578125  0.00689697]\n",
            "[-0.06542969 -0.09521484 -0.06225586  0.16210938  0.01989746]\n",
            "[-0.03198242  0.27148438 -0.2890625  -0.15429688  0.16894531]\n",
            "[ 0.11132812  0.10595703 -0.07373047  0.18847656  0.07666016]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "ac8b42811c924e7988f17b9dbd3f71ef",
          "grade": false,
          "grade_id": "cell-4ad44071d3785409",
          "locked": true,
          "schema_version": 1,
          "solution": false
        },
        "id": "2UBnMwiXKDVy"
      },
      "source": [
        "**Q3 (1 point)**: Find the euclidean distances between the word pairs : \n",
        "\n",
        "- (France, England)\n",
        "- (smaller, bigger)\n",
        "- (England, London)\n",
        "- (France, Rocket)\n",
        "- (big, bigger)\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "nbgrader": {
          "checksum": "a771483fbb59086604eb84bcc7c7f0ad",
          "grade": false,
          "grade_id": "cell-3aba86afc0ebd8a8",
          "locked": false,
          "schema_version": 1,
          "solution": true
        },
        "id": "zQGd-YVoKDV3"
      },
      "source": [
        "#Replace 0 with the code / value; Do not delete this cell\n",
        "eu_dist1 = np.linalg.norm(model['France']-model['England'])\n",
        "eu_dist2 = np.linalg.norm(model['smaller']-model['bigger'])\n",
        "eu_dist3 = np.linalg.norm(model['England']-model['London'])\n",
        "eu_dist4 = np.linalg.norm(model['France']-model['Rocket'])\n",
        "eu_dist5 = np.linalg.norm(model['big']-model['bigger'])\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "17796eb5de342e8f8e841aa137a2c41c",
          "grade": true,
          "grade_id": "cell-15ffa50b82de21ad",
          "locked": true,
          "points": 1,
          "schema_version": 1,
          "solution": false
        },
        "id": "HsSg0l2UKDV6",
        "outputId": "42efe9f2-52cc-4baf-b234-2187366ed5f6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 102
        }
      },
      "source": [
        "#This is an autograded cell, do not edit / delete\n",
        "print(eu_dist1)\n",
        "print(eu_dist2)\n",
        "print(eu_dist3)\n",
        "print(eu_dist4)\n",
        "print(eu_dist5)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "3.0151067\n",
            "1.8618743\n",
            "2.8752837\n",
            "3.892071\n",
            "1.9586496\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "bc50385c084e87555d0fa3e25c71bcea",
          "grade": false,
          "grade_id": "cell-7cd00f3c129dfab5",
          "locked": true,
          "schema_version": 1,
          "solution": false
        },
        "id": "OFll-_W3KDV8"
      },
      "source": [
        "**Q4 (2 points)**: What is the relationship between the magnitude of individual vectors, the vectors themselves and the cosine distance for any pair of words. Use any tuple in Q1 as an example to support your answer.\n",
        "\n",
        "**Do not delete the below cell**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "nbgrader": {
          "checksum": "f5858182370c73645303d9ce8103bbe2",
          "grade": true,
          "grade_id": "cell-09afc950f748a302",
          "locked": false,
          "points": 2,
          "schema_version": 1,
          "solution": true
        },
        "id": "rPUUyypZKDV9",
        "outputId": "c1b13d3a-0284-4d4b-9040-1d8ca1d0fbb6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# YOUR CODE HERE\n",
        "sim = np.dot(model['France'],model['England']) / (np.linalg.norm(model['France']) * np.linalg.norm(model['England']))\n",
        "sim\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.39804947"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CFxUmCfR1Gyv"
      },
      "source": [
        "A vector is represented by its magnitutde and it angle. If we look at France and Rocket as an example, we can see that their cosine distance is very small while their euclidean distance is large."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "afc0e843c7545e2df83448feda9f28f5",
          "grade": false,
          "grade_id": "cell-7cd8b9b67386376d",
          "locked": true,
          "schema_version": 1,
          "solution": false
        },
        "id": "XvO2iU7QKDWA"
      },
      "source": [
        "**Q5 (1 point)**: Time to dabble with the power of Word2Vec. Find the 2 closest words  for the following conditions:  \n",
        "- (King - Man + Queen)\n",
        "- (bigger - big + small)\n",
        "- (man + programmer - woman)\n",
        "- (waiting - wait + run)\n",
        "- (Texas + Milwaukee – Wisconsin)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "nbgrader": {
          "checksum": "50ef096feb166865434fe2fca3d41f99",
          "grade": false,
          "grade_id": "cell-b72201968c5fd1ec",
          "locked": false,
          "schema_version": 1,
          "solution": true
        },
        "id": "jCxWmA1eKDWB",
        "outputId": "82bf6785-5344-4a96-b855-381ec5c3bddd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        }
      },
      "source": [
        "#Replace 0 with the code / value; Do not delete this cell\n",
        "closest1 = model.most_similar(positive=['Queen', 'King'], negative=['Man'])[:2]\n",
        "closest2 = model.most_similar(positive=['bigger', 'small'], negative=['big'])[:2]\n",
        "closest3 = model.most_similar(positive=['man', 'programmer'], negative=['woman'])[:2]\n",
        "closest4 = model.most_similar(positive=['waiting', 'run'], negative=['wait'])[:2]\n",
        "closest5 = model.most_similar(positive=['Texas', 'Milwaukee'], negative=['Wisconsin'])[:2]\n",
        "\n",
        "\n",
        "\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/gensim/matutils.py:737: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
            "  if np.issubdtype(vec.dtype, np.int):\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "f9c5ff502264f29d2632c6387f92686a",
          "grade": true,
          "grade_id": "cell-b69718ab0e1470bc",
          "locked": true,
          "points": 1,
          "schema_version": 1,
          "solution": false
        },
        "id": "io9elfD8KDWE",
        "outputId": "f2fa19fc-a23f-4326-d83e-bdee09aa3210",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 102
        }
      },
      "source": [
        "#This is an autograded cell, do not edit/delete\n",
        "print(closest1)\n",
        "print(closest2)\n",
        "print(closest3)\n",
        "print(closest4)\n",
        "print(closest5)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[('Queen_Elizabeth', 0.5257916450500488), ('monarch', 0.5004087090492249)]\n",
            "[('larger', 0.7402471899986267), ('smaller', 0.732999324798584)]\n",
            "[('programer', 0.5371963977813721), ('programmers', 0.5310999155044556)]\n",
            "[('running', 0.5654535889625549), ('runs', 0.49640005826950073)]\n",
            "[('Houston', 0.7767744064331055), ('Fort_Worth', 0.7270511388778687)]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "6432058d78f4fa52224c48a3b3e71d0d",
          "grade": false,
          "grade_id": "cell-73dca0e2072fef91",
          "locked": true,
          "schema_version": 1,
          "solution": false
        },
        "id": "erUu4u71KDWJ"
      },
      "source": [
        "**Q6 (3 points)**: Using the vectors for the words in the Google News dataset, explore the semantic representation of these words through K-means clustering and explain your findings.\n",
        "\n",
        "*Note : Since there are ~3Mil words in the vocabulary, you can downsample it to ~20-30k randomly selected words*\n",
        "\n",
        "**Do not delete the below cell**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "nbgrader": {
          "checksum": "7ecef46689f11d4d0a6fed72e049235f",
          "grade": true,
          "grade_id": "cell-80b177848b8b0212",
          "locked": false,
          "points": 3,
          "schema_version": 1,
          "solution": true
        },
        "id": "M3jN02fOKDWK",
        "outputId": "47dcc1b5-1a8b-43c6-ce37-c28ba32d1764",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 253
        }
      },
      "source": [
        "# YOUR CODE HERE\n",
        "from sklearn.cluster import KMeans\n",
        "import random\n",
        "random_keys = random.sample(list(model.vocab), 20000)\n",
        "random_vectors = model[random_keys]\n",
        "df = pd.DataFrame(random_vectors)\n",
        "df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "      <th>2</th>\n",
              "      <th>3</th>\n",
              "      <th>4</th>\n",
              "      <th>5</th>\n",
              "      <th>6</th>\n",
              "      <th>7</th>\n",
              "      <th>8</th>\n",
              "      <th>9</th>\n",
              "      <th>10</th>\n",
              "      <th>11</th>\n",
              "      <th>12</th>\n",
              "      <th>13</th>\n",
              "      <th>14</th>\n",
              "      <th>15</th>\n",
              "      <th>16</th>\n",
              "      <th>17</th>\n",
              "      <th>18</th>\n",
              "      <th>19</th>\n",
              "      <th>20</th>\n",
              "      <th>21</th>\n",
              "      <th>22</th>\n",
              "      <th>23</th>\n",
              "      <th>24</th>\n",
              "      <th>25</th>\n",
              "      <th>26</th>\n",
              "      <th>27</th>\n",
              "      <th>28</th>\n",
              "      <th>29</th>\n",
              "      <th>30</th>\n",
              "      <th>31</th>\n",
              "      <th>32</th>\n",
              "      <th>33</th>\n",
              "      <th>34</th>\n",
              "      <th>35</th>\n",
              "      <th>36</th>\n",
              "      <th>37</th>\n",
              "      <th>38</th>\n",
              "      <th>39</th>\n",
              "      <th>...</th>\n",
              "      <th>260</th>\n",
              "      <th>261</th>\n",
              "      <th>262</th>\n",
              "      <th>263</th>\n",
              "      <th>264</th>\n",
              "      <th>265</th>\n",
              "      <th>266</th>\n",
              "      <th>267</th>\n",
              "      <th>268</th>\n",
              "      <th>269</th>\n",
              "      <th>270</th>\n",
              "      <th>271</th>\n",
              "      <th>272</th>\n",
              "      <th>273</th>\n",
              "      <th>274</th>\n",
              "      <th>275</th>\n",
              "      <th>276</th>\n",
              "      <th>277</th>\n",
              "      <th>278</th>\n",
              "      <th>279</th>\n",
              "      <th>280</th>\n",
              "      <th>281</th>\n",
              "      <th>282</th>\n",
              "      <th>283</th>\n",
              "      <th>284</th>\n",
              "      <th>285</th>\n",
              "      <th>286</th>\n",
              "      <th>287</th>\n",
              "      <th>288</th>\n",
              "      <th>289</th>\n",
              "      <th>290</th>\n",
              "      <th>291</th>\n",
              "      <th>292</th>\n",
              "      <th>293</th>\n",
              "      <th>294</th>\n",
              "      <th>295</th>\n",
              "      <th>296</th>\n",
              "      <th>297</th>\n",
              "      <th>298</th>\n",
              "      <th>299</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.016235</td>\n",
              "      <td>-0.000999</td>\n",
              "      <td>-0.023071</td>\n",
              "      <td>0.009460</td>\n",
              "      <td>0.005615</td>\n",
              "      <td>-0.050781</td>\n",
              "      <td>-0.061035</td>\n",
              "      <td>-0.055908</td>\n",
              "      <td>0.181641</td>\n",
              "      <td>0.038330</td>\n",
              "      <td>-0.013489</td>\n",
              "      <td>0.096191</td>\n",
              "      <td>0.025391</td>\n",
              "      <td>0.009644</td>\n",
              "      <td>0.083984</td>\n",
              "      <td>-0.017944</td>\n",
              "      <td>0.029785</td>\n",
              "      <td>0.058350</td>\n",
              "      <td>0.033936</td>\n",
              "      <td>0.046875</td>\n",
              "      <td>-0.019653</td>\n",
              "      <td>0.020020</td>\n",
              "      <td>-0.072266</td>\n",
              "      <td>-0.050537</td>\n",
              "      <td>0.001228</td>\n",
              "      <td>-0.029175</td>\n",
              "      <td>-0.024902</td>\n",
              "      <td>0.029663</td>\n",
              "      <td>0.007568</td>\n",
              "      <td>0.036377</td>\n",
              "      <td>0.095703</td>\n",
              "      <td>0.046631</td>\n",
              "      <td>0.005035</td>\n",
              "      <td>-0.052246</td>\n",
              "      <td>0.002228</td>\n",
              "      <td>0.051270</td>\n",
              "      <td>-0.033691</td>\n",
              "      <td>-0.033691</td>\n",
              "      <td>0.014832</td>\n",
              "      <td>0.069824</td>\n",
              "      <td>...</td>\n",
              "      <td>0.101562</td>\n",
              "      <td>0.072754</td>\n",
              "      <td>0.092285</td>\n",
              "      <td>0.024048</td>\n",
              "      <td>0.082520</td>\n",
              "      <td>0.055176</td>\n",
              "      <td>0.009155</td>\n",
              "      <td>0.023193</td>\n",
              "      <td>0.076172</td>\n",
              "      <td>-0.059814</td>\n",
              "      <td>-0.052246</td>\n",
              "      <td>-0.069824</td>\n",
              "      <td>-0.072754</td>\n",
              "      <td>-0.052490</td>\n",
              "      <td>0.038574</td>\n",
              "      <td>0.022217</td>\n",
              "      <td>-0.045166</td>\n",
              "      <td>0.054932</td>\n",
              "      <td>-0.058105</td>\n",
              "      <td>-0.082520</td>\n",
              "      <td>-0.059326</td>\n",
              "      <td>-0.059326</td>\n",
              "      <td>-0.064453</td>\n",
              "      <td>0.063965</td>\n",
              "      <td>-0.035400</td>\n",
              "      <td>0.070312</td>\n",
              "      <td>-0.086426</td>\n",
              "      <td>0.113770</td>\n",
              "      <td>-0.000874</td>\n",
              "      <td>-0.000257</td>\n",
              "      <td>-0.006439</td>\n",
              "      <td>-0.026978</td>\n",
              "      <td>-0.003723</td>\n",
              "      <td>0.063477</td>\n",
              "      <td>0.007996</td>\n",
              "      <td>-0.035156</td>\n",
              "      <td>-0.019775</td>\n",
              "      <td>-0.002090</td>\n",
              "      <td>0.024780</td>\n",
              "      <td>0.042480</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.162109</td>\n",
              "      <td>-0.226562</td>\n",
              "      <td>0.013428</td>\n",
              "      <td>-0.085449</td>\n",
              "      <td>0.029785</td>\n",
              "      <td>-0.196289</td>\n",
              "      <td>-0.220703</td>\n",
              "      <td>-0.103516</td>\n",
              "      <td>0.087402</td>\n",
              "      <td>0.092773</td>\n",
              "      <td>0.171875</td>\n",
              "      <td>-0.168945</td>\n",
              "      <td>0.110352</td>\n",
              "      <td>-0.125000</td>\n",
              "      <td>-0.082031</td>\n",
              "      <td>0.013306</td>\n",
              "      <td>0.092773</td>\n",
              "      <td>0.179688</td>\n",
              "      <td>0.083496</td>\n",
              "      <td>0.074707</td>\n",
              "      <td>-0.000234</td>\n",
              "      <td>-0.065918</td>\n",
              "      <td>0.078613</td>\n",
              "      <td>-0.004028</td>\n",
              "      <td>0.167969</td>\n",
              "      <td>-0.154297</td>\n",
              "      <td>0.142578</td>\n",
              "      <td>-0.009460</td>\n",
              "      <td>0.144531</td>\n",
              "      <td>-0.212891</td>\n",
              "      <td>0.019653</td>\n",
              "      <td>-0.126953</td>\n",
              "      <td>-0.095703</td>\n",
              "      <td>-0.077637</td>\n",
              "      <td>0.003326</td>\n",
              "      <td>-0.043701</td>\n",
              "      <td>-0.191406</td>\n",
              "      <td>-0.056396</td>\n",
              "      <td>-0.091797</td>\n",
              "      <td>0.041748</td>\n",
              "      <td>...</td>\n",
              "      <td>0.181641</td>\n",
              "      <td>0.075684</td>\n",
              "      <td>-0.153320</td>\n",
              "      <td>-0.084961</td>\n",
              "      <td>-0.122070</td>\n",
              "      <td>-0.173828</td>\n",
              "      <td>-0.065918</td>\n",
              "      <td>0.058350</td>\n",
              "      <td>0.049805</td>\n",
              "      <td>-0.085938</td>\n",
              "      <td>0.151367</td>\n",
              "      <td>-0.005249</td>\n",
              "      <td>0.020020</td>\n",
              "      <td>0.013184</td>\n",
              "      <td>-0.022217</td>\n",
              "      <td>0.057617</td>\n",
              "      <td>-0.089844</td>\n",
              "      <td>0.014099</td>\n",
              "      <td>-0.031494</td>\n",
              "      <td>-0.034668</td>\n",
              "      <td>-0.027100</td>\n",
              "      <td>0.026489</td>\n",
              "      <td>-0.044678</td>\n",
              "      <td>-0.054932</td>\n",
              "      <td>-0.091309</td>\n",
              "      <td>-0.111328</td>\n",
              "      <td>-0.035156</td>\n",
              "      <td>0.127930</td>\n",
              "      <td>0.044189</td>\n",
              "      <td>0.233398</td>\n",
              "      <td>0.072754</td>\n",
              "      <td>-0.128906</td>\n",
              "      <td>0.117188</td>\n",
              "      <td>0.056885</td>\n",
              "      <td>-0.144531</td>\n",
              "      <td>-0.048828</td>\n",
              "      <td>-0.158203</td>\n",
              "      <td>0.018677</td>\n",
              "      <td>0.085938</td>\n",
              "      <td>0.052002</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.014587</td>\n",
              "      <td>-0.111328</td>\n",
              "      <td>-0.078613</td>\n",
              "      <td>0.032715</td>\n",
              "      <td>-0.096680</td>\n",
              "      <td>0.102539</td>\n",
              "      <td>0.007935</td>\n",
              "      <td>0.111328</td>\n",
              "      <td>0.027222</td>\n",
              "      <td>-0.095215</td>\n",
              "      <td>0.060303</td>\n",
              "      <td>-0.058350</td>\n",
              "      <td>-0.084473</td>\n",
              "      <td>0.202148</td>\n",
              "      <td>-0.040283</td>\n",
              "      <td>0.067383</td>\n",
              "      <td>0.062500</td>\n",
              "      <td>0.059814</td>\n",
              "      <td>-0.073242</td>\n",
              "      <td>-0.013062</td>\n",
              "      <td>0.031982</td>\n",
              "      <td>-0.149414</td>\n",
              "      <td>-0.121094</td>\n",
              "      <td>0.052490</td>\n",
              "      <td>-0.020630</td>\n",
              "      <td>-0.122559</td>\n",
              "      <td>-0.082031</td>\n",
              "      <td>0.050049</td>\n",
              "      <td>-0.053223</td>\n",
              "      <td>-0.040771</td>\n",
              "      <td>-0.021729</td>\n",
              "      <td>-0.083008</td>\n",
              "      <td>-0.025635</td>\n",
              "      <td>0.112305</td>\n",
              "      <td>0.018311</td>\n",
              "      <td>-0.160156</td>\n",
              "      <td>0.021606</td>\n",
              "      <td>0.006104</td>\n",
              "      <td>0.002686</td>\n",
              "      <td>0.047852</td>\n",
              "      <td>...</td>\n",
              "      <td>0.088867</td>\n",
              "      <td>-0.038086</td>\n",
              "      <td>-0.117676</td>\n",
              "      <td>-0.033936</td>\n",
              "      <td>0.053223</td>\n",
              "      <td>0.028076</td>\n",
              "      <td>-0.020996</td>\n",
              "      <td>0.011047</td>\n",
              "      <td>-0.046387</td>\n",
              "      <td>-0.119141</td>\n",
              "      <td>0.154297</td>\n",
              "      <td>0.068848</td>\n",
              "      <td>0.083008</td>\n",
              "      <td>0.196289</td>\n",
              "      <td>0.094238</td>\n",
              "      <td>-0.075684</td>\n",
              "      <td>0.064941</td>\n",
              "      <td>-0.151367</td>\n",
              "      <td>-0.132812</td>\n",
              "      <td>-0.020752</td>\n",
              "      <td>-0.078613</td>\n",
              "      <td>-0.065430</td>\n",
              "      <td>0.048096</td>\n",
              "      <td>0.147461</td>\n",
              "      <td>0.013123</td>\n",
              "      <td>-0.010193</td>\n",
              "      <td>0.004120</td>\n",
              "      <td>0.063965</td>\n",
              "      <td>0.072266</td>\n",
              "      <td>0.043213</td>\n",
              "      <td>0.004150</td>\n",
              "      <td>0.052490</td>\n",
              "      <td>-0.061279</td>\n",
              "      <td>-0.012329</td>\n",
              "      <td>-0.055420</td>\n",
              "      <td>0.059814</td>\n",
              "      <td>0.062256</td>\n",
              "      <td>-0.251953</td>\n",
              "      <td>-0.144531</td>\n",
              "      <td>-0.046631</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>-0.030884</td>\n",
              "      <td>-0.028931</td>\n",
              "      <td>0.086426</td>\n",
              "      <td>0.008484</td>\n",
              "      <td>0.062988</td>\n",
              "      <td>0.017822</td>\n",
              "      <td>0.024292</td>\n",
              "      <td>-0.034180</td>\n",
              "      <td>0.014709</td>\n",
              "      <td>0.064453</td>\n",
              "      <td>-0.040039</td>\n",
              "      <td>0.020386</td>\n",
              "      <td>0.059082</td>\n",
              "      <td>0.077637</td>\n",
              "      <td>-0.105957</td>\n",
              "      <td>0.177734</td>\n",
              "      <td>0.052979</td>\n",
              "      <td>-0.012207</td>\n",
              "      <td>-0.020752</td>\n",
              "      <td>-0.025146</td>\n",
              "      <td>-0.031738</td>\n",
              "      <td>-0.066406</td>\n",
              "      <td>-0.048096</td>\n",
              "      <td>0.070801</td>\n",
              "      <td>0.069336</td>\n",
              "      <td>-0.056885</td>\n",
              "      <td>0.025391</td>\n",
              "      <td>0.038330</td>\n",
              "      <td>-0.098145</td>\n",
              "      <td>0.051514</td>\n",
              "      <td>0.072266</td>\n",
              "      <td>-0.008484</td>\n",
              "      <td>-0.139648</td>\n",
              "      <td>0.060303</td>\n",
              "      <td>0.033203</td>\n",
              "      <td>-0.011169</td>\n",
              "      <td>0.008240</td>\n",
              "      <td>0.096191</td>\n",
              "      <td>-0.041504</td>\n",
              "      <td>-0.036133</td>\n",
              "      <td>...</td>\n",
              "      <td>-0.085449</td>\n",
              "      <td>0.034180</td>\n",
              "      <td>-0.061523</td>\n",
              "      <td>-0.015137</td>\n",
              "      <td>0.065918</td>\n",
              "      <td>0.072266</td>\n",
              "      <td>0.002609</td>\n",
              "      <td>0.039062</td>\n",
              "      <td>0.019531</td>\n",
              "      <td>-0.045898</td>\n",
              "      <td>0.086914</td>\n",
              "      <td>0.059814</td>\n",
              "      <td>0.018677</td>\n",
              "      <td>0.026245</td>\n",
              "      <td>-0.016235</td>\n",
              "      <td>-0.024536</td>\n",
              "      <td>0.013794</td>\n",
              "      <td>-0.126953</td>\n",
              "      <td>-0.023560</td>\n",
              "      <td>0.028198</td>\n",
              "      <td>0.031250</td>\n",
              "      <td>-0.070801</td>\n",
              "      <td>-0.028931</td>\n",
              "      <td>-0.051270</td>\n",
              "      <td>-0.146484</td>\n",
              "      <td>-0.003616</td>\n",
              "      <td>0.040283</td>\n",
              "      <td>0.043213</td>\n",
              "      <td>-0.007111</td>\n",
              "      <td>0.137695</td>\n",
              "      <td>-0.057373</td>\n",
              "      <td>0.078613</td>\n",
              "      <td>-0.008301</td>\n",
              "      <td>0.070801</td>\n",
              "      <td>-0.133789</td>\n",
              "      <td>0.023438</td>\n",
              "      <td>0.004883</td>\n",
              "      <td>-0.141602</td>\n",
              "      <td>-0.053223</td>\n",
              "      <td>0.059570</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>-0.038574</td>\n",
              "      <td>0.033203</td>\n",
              "      <td>0.033447</td>\n",
              "      <td>0.076172</td>\n",
              "      <td>-0.085449</td>\n",
              "      <td>-0.015991</td>\n",
              "      <td>-0.104004</td>\n",
              "      <td>0.046143</td>\n",
              "      <td>0.003601</td>\n",
              "      <td>0.090820</td>\n",
              "      <td>-0.003296</td>\n",
              "      <td>0.062988</td>\n",
              "      <td>0.032715</td>\n",
              "      <td>-0.024048</td>\n",
              "      <td>-0.008606</td>\n",
              "      <td>-0.033936</td>\n",
              "      <td>-0.042236</td>\n",
              "      <td>0.021118</td>\n",
              "      <td>-0.041504</td>\n",
              "      <td>-0.108887</td>\n",
              "      <td>-0.006317</td>\n",
              "      <td>0.010376</td>\n",
              "      <td>0.013245</td>\n",
              "      <td>0.099121</td>\n",
              "      <td>-0.016846</td>\n",
              "      <td>-0.022827</td>\n",
              "      <td>-0.071777</td>\n",
              "      <td>0.026733</td>\n",
              "      <td>0.015564</td>\n",
              "      <td>-0.067383</td>\n",
              "      <td>-0.037598</td>\n",
              "      <td>0.093750</td>\n",
              "      <td>0.056641</td>\n",
              "      <td>-0.035645</td>\n",
              "      <td>0.009277</td>\n",
              "      <td>0.004456</td>\n",
              "      <td>0.012268</td>\n",
              "      <td>0.061523</td>\n",
              "      <td>0.126953</td>\n",
              "      <td>-0.096680</td>\n",
              "      <td>...</td>\n",
              "      <td>0.045898</td>\n",
              "      <td>0.003937</td>\n",
              "      <td>-0.076660</td>\n",
              "      <td>-0.048340</td>\n",
              "      <td>-0.058838</td>\n",
              "      <td>0.110352</td>\n",
              "      <td>-0.049072</td>\n",
              "      <td>-0.097168</td>\n",
              "      <td>0.047607</td>\n",
              "      <td>0.018677</td>\n",
              "      <td>-0.082031</td>\n",
              "      <td>0.015869</td>\n",
              "      <td>-0.101074</td>\n",
              "      <td>0.008301</td>\n",
              "      <td>0.002197</td>\n",
              "      <td>0.141602</td>\n",
              "      <td>-0.072266</td>\n",
              "      <td>0.029663</td>\n",
              "      <td>0.014832</td>\n",
              "      <td>0.030029</td>\n",
              "      <td>-0.063477</td>\n",
              "      <td>0.014221</td>\n",
              "      <td>0.042725</td>\n",
              "      <td>0.029541</td>\n",
              "      <td>-0.041748</td>\n",
              "      <td>-0.072266</td>\n",
              "      <td>0.004456</td>\n",
              "      <td>0.006927</td>\n",
              "      <td>0.009460</td>\n",
              "      <td>0.053711</td>\n",
              "      <td>-0.057617</td>\n",
              "      <td>0.025757</td>\n",
              "      <td>-0.106445</td>\n",
              "      <td>-0.089844</td>\n",
              "      <td>-0.005249</td>\n",
              "      <td>0.004181</td>\n",
              "      <td>0.019653</td>\n",
              "      <td>-0.009033</td>\n",
              "      <td>0.072266</td>\n",
              "      <td>0.005371</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>5 rows × 300 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "        0         1         2    ...       297       298       299\n",
              "0  0.016235 -0.000999 -0.023071  ... -0.002090  0.024780  0.042480\n",
              "1  0.162109 -0.226562  0.013428  ...  0.018677  0.085938  0.052002\n",
              "2  0.014587 -0.111328 -0.078613  ... -0.251953 -0.144531 -0.046631\n",
              "3 -0.030884 -0.028931  0.086426  ... -0.141602 -0.053223  0.059570\n",
              "4 -0.038574  0.033203  0.033447  ... -0.009033  0.072266  0.005371\n",
              "\n",
              "[5 rows x 300 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_p6BO-PL9mUo",
        "outputId": "10dc318b-ce1b-40a6-9aff-ea35c184b2f6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# from sklearn.metrics import silhouette_score\n",
        "# k_choices = [2, 3, 4, 5, 6, 7, 8]\n",
        "# scores = []\n",
        "# X = df\n",
        "# for k in k_choices:\n",
        "#     kmeans = KMeans(n_clusters=k, random_state=42)\n",
        "#     kmeans.fit(X)\n",
        "#     score = silhouette_score(X, kmeans.labels_)\n",
        "#     scores.append(score)\n",
        "# k_choices[scores.index(max(scores))]\n",
        "print(2) #takes too much time to run\n",
        "X = df"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "C5rclZy2_eKS",
        "outputId": "cda5f374-5cc7-433b-bc3f-044805025931",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "from collections import Counter, defaultdict\n",
        "kmeans = KMeans(n_clusters=2, random_state=42)\n",
        "kmeans.fit(X)\n",
        "Counter(kmeans.labels_)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Counter({0: 4359, 1: 15641})"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7jzKxME89mAS",
        "outputId": "4ca41607-d36a-41f8-e398-2a0e66067219",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "kmeans.labels_"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([1, 1, 1, ..., 1, 1, 1], dtype=int32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c-kG908f-i6j",
        "outputId": "e126815c-28ed-4d46-8398-b94733290981",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        }
      },
      "source": [
        "cluster = pd.DataFrame({'keys': random_keys, 'Cluster':kmeans.labels_})\n",
        "cluster"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>keys</th>\n",
              "      <th>Cluster</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Vean_Gregg</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Bali_Denpasar_airport</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>redact_sensitive</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Judge_Davinder_Lachlar</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>lipless_cranks</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19995</th>\n",
              "      <td>ADOT_advises</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19996</th>\n",
              "      <td>Franziska_Achterberg</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19997</th>\n",
              "      <td>Pasat</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19998</th>\n",
              "      <td>painstakingly_hoisted</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19999</th>\n",
              "      <td>Shinns</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>20000 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                         keys  Cluster\n",
              "0                  Vean_Gregg        1\n",
              "1       Bali_Denpasar_airport        1\n",
              "2            redact_sensitive        1\n",
              "3      Judge_Davinder_Lachlar        1\n",
              "4              lipless_cranks        1\n",
              "...                       ...      ...\n",
              "19995            ADOT_advises        1\n",
              "19996    Franziska_Achterberg        1\n",
              "19997                   Pasat        1\n",
              "19998   painstakingly_hoisted        1\n",
              "19999                  Shinns        1\n",
              "\n",
              "[20000 rows x 2 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1FKqU7FF_Ei3",
        "outputId": "8f02180f-b8eb-4061-a747-97e9878f196b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        }
      },
      "source": [
        "zeros = cluster[cluster['Cluster']==0]\n",
        "zeros.iloc[:90]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>keys</th>\n",
              "      <th>Cluster</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>bioactive_lipid</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>Universal_Caller_ID</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>Mars_Needs_Moms</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>Utatu</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>brothy</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>392</th>\n",
              "      <td>www.merge.com</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>394</th>\n",
              "      <td>IntraLase_Corp.</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>397</th>\n",
              "      <td>markets_COLAZAL</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>402</th>\n",
              "      <td>Oglebay_Norton</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>410</th>\n",
              "      <td>AVCHD_Lite</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>90 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                    keys  Cluster\n",
              "5        bioactive_lipid        0\n",
              "7    Universal_Caller_ID        0\n",
              "10       Mars_Needs_Moms        0\n",
              "11                 Utatu        0\n",
              "13                brothy        0\n",
              "..                   ...      ...\n",
              "392        www.merge.com        0\n",
              "394      IntraLase_Corp.        0\n",
              "397      markets_COLAZAL        0\n",
              "402       Oglebay_Norton        0\n",
              "410           AVCHD_Lite        0\n",
              "\n",
              "[90 rows x 2 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AYAcXaiJ_YzH",
        "outputId": "302a0d2f-11df-4d12-d5e8-d220cb520229",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        }
      },
      "source": [
        "ones = cluster[cluster['Cluster']==1]\n",
        "ones.iloc[:150]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>keys</th>\n",
              "      <th>Cluster</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Vean_Gregg</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Bali_Denpasar_airport</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>redact_sensitive</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Judge_Davinder_Lachlar</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>lipless_cranks</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>189</th>\n",
              "      <td>wheeled_tuk_tuk</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>190</th>\n",
              "      <td>Kabbalist</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>191</th>\n",
              "      <td>Hermen_Shastri</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>194</th>\n",
              "      <td>wah_wah_guitars</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>195</th>\n",
              "      <td>calvers</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>150 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                       keys  Cluster\n",
              "0                Vean_Gregg        1\n",
              "1     Bali_Denpasar_airport        1\n",
              "2          redact_sensitive        1\n",
              "3    Judge_Davinder_Lachlar        1\n",
              "4            lipless_cranks        1\n",
              "..                      ...      ...\n",
              "189         wheeled_tuk_tuk        1\n",
              "190               Kabbalist        1\n",
              "191          Hermen_Shastri        1\n",
              "194         wah_wah_guitars        1\n",
              "195                 calvers        1\n",
              "\n",
              "[150 rows x 2 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "y_wIfTsU_w6l"
      },
      "source": [
        "From my analysis, I decided that it is best to have 2 clusters from my trial from 2 to 8. There seems to be much more points classified as cluster 0 in comparison to cluster 1. Cluster 0 seems to have more negative words such as pistols or shootings. However, both clusters show a large amount of proper nouns which makes it hard to determine what is considered a group by eye and interpretation."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "0467b27a0f59504cbb62b851a002386f",
          "grade": false,
          "grade_id": "cell-5b2a5e8ff6c74323",
          "locked": true,
          "schema_version": 1,
          "solution": false
        },
        "id": "rmdtLoHkKDWR"
      },
      "source": [
        "**Q7 (1 point)**: What loss function does the skipgram model use and briefly describe what this function is minimizing.\n",
        "\n",
        "**Do not delete the below cell**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "nbgrader": {
          "checksum": "774aef2c5bf8ef9d92e3489d1cd80390",
          "grade": true,
          "grade_id": "cell-90cc4b2c0ae8e2c2",
          "locked": false,
          "points": 1,
          "schema_version": 1,
          "solution": true
        },
        "id": "SyOASYXOKDWS"
      },
      "source": [
        "# YOUR CODE HERE\n",
        "answer = \"Skipgram uses softmax loss function. We try to minimize the cross entropy between the estimated class probability and the true dis\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "deletable": false,
        "editable": false,
        "nbgrader": {
          "checksum": "c14f6069f64cc86ab6e384d28df270d8",
          "grade": false,
          "grade_id": "cell-74a177caaabb5009",
          "locked": true,
          "schema_version": 1,
          "solution": false
        },
        "id": "dbpuJx9CKDWV"
      },
      "source": [
        "**Bonus Question (1 point)** : Find at least 2 interesting word vec combinations like the ones given in Q5\n",
        "\n",
        "**Do not delete the below cell**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "deletable": false,
        "nbgrader": {
          "checksum": "c2d42b5327f4b020c7e1706506dd5ce9",
          "grade": true,
          "grade_id": "cell-7351297993d72e83",
          "locked": false,
          "points": 1,
          "schema_version": 1,
          "solution": true
        },
        "id": "pQM8C_T7KDWW",
        "outputId": "9e74d573-64b0-48c1-ba0d-a3d50ff1cebe",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "# YOUR CODE HERE\n",
        "cool1 = model.most_similar(positive=['Fast', 'Small'], negative=['Slow'])[:2]\n",
        "cool2 = model.most_similar(positive=['Happiness', 'Sadness'], negative=['Happy'])[:2]\n",
        "print(cool1), print(cool2)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/gensim/matutils.py:737: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
            "  if np.issubdtype(vec.dtype, np.int):\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "[('small', 0.4256536662578583), ('Large', 0.40149611234664917)]\n",
            "[('Loneliness', 0.4631291627883911), ('Emptiness', 0.45362788438796997)]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(None, None)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LxmddAQq8MPw"
      },
      "source": [
        "# Part 2"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ql_K6ulS1A4c"
      },
      "source": [
        "# 1. Embeddings"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R96BEqSC8P3T"
      },
      "source": [
        "url = \"http://www.gutenberg.org/files/25344/25344-0.txt\" \n",
        "resp = urlopen(url)\n",
        "raw = resp.read().decode('utf8')\n",
        "firstlook = tokenize.sent_tokenize(raw)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nd_noMFD8ajr"
      },
      "source": [
        "pattern = r'''(?x)  # set flag to allow verbose regexps\n",
        "(?:[A-Z]\\.)+        # abbreviations, e.g. U.S.A.\n",
        "|\\w+(?:[-']\\w+)*    # words with optional internal hyphens\n",
        "|\\$?\\d+(?:\\.\\d+)?   # currency, e.g. $12.80 \n",
        "|\\.\\.\\.             # elipses\n",
        "|[.,;\"'?()-_`]      # these are separate tokens\n",
        "'''\n",
        "tokenized_raw = \" \".join(nltk.regexp_tokenize(raw, pattern))\n",
        "tokenized_raw = tokenize.sent_tokenize(tokenized_raw)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "epM8TzYO8c1I",
        "outputId": "e689e182-77a5-4b32-c47e-83dca5a84ad7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 207
        }
      },
      "source": [
        "tokenized_raw[:10]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['The Project Gutenberg EBook of The Scarlet Letter , by Nathaniel Hawthorne This eBook is for the use of anyone anywhere at no cost and with almost no restrictions whatsoever .',\n",
              " 'You may copy it , give it away or re-use it under the terms of the Project Gutenberg License included with this eBook or online at www .',\n",
              " 'gutenberg .',\n",
              " 'net Title : The Scarlet Letter Author : Nathaniel Hawthorne Illustrator : Mary Hallock Foote L. S. Ipsen Release Date : May 5 , 2008 [ EBook 25344 ] Language : English Character set encoding : UTF-8 * * * START OF THIS PROJECT GUTENBERG EBOOK THE SCARLET LETTER * * * Produced by Markus Brenner , Irma Spehar and the Online Distributed Proofreading Team at http : / / www .',\n",
              " 'pgdp .',\n",
              " 'net ( This file was produced from images generously made available by The Internet Archive / American Libraries . )',\n",
              " 'THE SCARLET LETTER .',\n",
              " 'BY NATHANIEL HAWTHORNE .',\n",
              " 'Illustrated .',\n",
              " '[ Illustration ] BOSTON : JAMES R. OSGOOD AND COMPANY , LATE TICKNOR FIELDS , AND FIELDS , OSGOOD , CO .']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "icaGNaAC8lQv"
      },
      "source": [
        "# Remove punctuations\n",
        "nopunct = []\n",
        "for sent in tokenized_raw:\n",
        "    a = [w for w in sent.split() if w not in string.punctuation]\n",
        "    nopunct.append(\" \".join(a))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jd1yuyzZ8qZW",
        "outputId": "1107ba9a-7d38-4926-ee68-e1dc771f1400",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 969
        }
      },
      "source": [
        "# Word tokenize\n",
        "tok_corp = [nltk.word_tokenize(sent) for sent in nopunct]\n",
        "tok_corp[:3]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[['The',\n",
              "  'Project',\n",
              "  'Gutenberg',\n",
              "  'EBook',\n",
              "  'of',\n",
              "  'The',\n",
              "  'Scarlet',\n",
              "  'Letter',\n",
              "  'by',\n",
              "  'Nathaniel',\n",
              "  'Hawthorne',\n",
              "  'This',\n",
              "  'eBook',\n",
              "  'is',\n",
              "  'for',\n",
              "  'the',\n",
              "  'use',\n",
              "  'of',\n",
              "  'anyone',\n",
              "  'anywhere',\n",
              "  'at',\n",
              "  'no',\n",
              "  'cost',\n",
              "  'and',\n",
              "  'with',\n",
              "  'almost',\n",
              "  'no',\n",
              "  'restrictions',\n",
              "  'whatsoever'],\n",
              " ['You',\n",
              "  'may',\n",
              "  'copy',\n",
              "  'it',\n",
              "  'give',\n",
              "  'it',\n",
              "  'away',\n",
              "  'or',\n",
              "  're-use',\n",
              "  'it',\n",
              "  'under',\n",
              "  'the',\n",
              "  'terms',\n",
              "  'of',\n",
              "  'the',\n",
              "  'Project',\n",
              "  'Gutenberg',\n",
              "  'License',\n",
              "  'included',\n",
              "  'with',\n",
              "  'this',\n",
              "  'eBook',\n",
              "  'or',\n",
              "  'online',\n",
              "  'at',\n",
              "  'www'],\n",
              " ['gutenberg']]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zJ6lVmVp8tUQ",
        "outputId": "85ab7c20-6084-4a88-c8a8-ddc85b6987eb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "### Creating a list of unique words \n",
        "unique_words = list(set([item for sublist in tok_corp for item in sublist]))\n",
        "len(unique_words)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "9932"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SfLu9Rjn80Tm",
        "outputId": "6f3c3421-516f-436f-9da2-645b9cade46b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "#pretrained model\n",
        "model"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<gensim.models.keyedvectors.Word2VecKeyedVectors at 0x7f0b998e9a58>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Guy1K0958_6B",
        "outputId": "fa64a919-be36-4d47-8802-9eb2a976203f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        }
      },
      "source": [
        "## Sample for similarity \n",
        "print (model.most_similar(positive=['king'], negative=['man']))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/gensim/matutils.py:737: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
            "  if np.issubdtype(vec.dtype, np.int):\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "[('kings', 0.4295138418674469), ('queen', 0.39028695225715637), ('Pansy_Ho_Chiu', 0.3827225863933563), ('monarch', 0.3633837103843689), ('kingdom', 0.36145076155662537), ('royal_palace', 0.3535977602005005), ('Savory_aromas_wafted', 0.35272473096847534), ('princes', 0.3526379466056824), ('monarchy', 0.3432357907295227), ('Rama_VII', 0.3380342423915863)]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U6apUFDy9Efh"
      },
      "source": [
        "model_b = gensim.models.Word2Vec(tok_corp, min_count=1, size=16, window=5)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-F67mesl9MO4",
        "outputId": "6d19a1e6-e803-4099-d2b7-ccd05931c262",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 292
        }
      },
      "source": [
        "## Extracting the respective vectors corresponding to the words\n",
        "vector_list = model_b[unique_words] ## n by d matrix containing words and their respective vectors\n",
        "vector_list"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:1: DeprecationWarning: Call to deprecated `__getitem__` (Method will be removed in 4.0.0, use self.wv.__getitem__() instead).\n",
            "  \"\"\"Entry point for launching an IPython kernel.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[ 0.01543421,  0.035304  , -0.01397987, ...,  0.05129982,\n",
              "        -0.05719861, -0.00063438],\n",
              "       [ 0.02713646,  0.03406958, -0.01241592, ...,  0.04459162,\n",
              "        -0.00765951,  0.0191385 ],\n",
              "       [ 0.00677163, -0.00337087,  0.02717992, ...,  0.02137208,\n",
              "        -0.07194926, -0.00120875],\n",
              "       ...,\n",
              "       [ 0.01868029,  0.05652009,  0.06032439, ...,  0.20363979,\n",
              "        -0.16454342,  0.08424284],\n",
              "       [-0.01430996,  0.02555185,  0.01940421, ...,  0.04508232,\n",
              "        -0.00822242,  0.01319215],\n",
              "       [-0.02335694,  0.0167746 , -0.00922698, ...,  0.03335363,\n",
              "        -0.02797239,  0.02146843]], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1bCg36yw9QcO",
        "outputId": "4ac9f9e9-2ed7-46d6-c437-9d6cd4b12727",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "len(vector_list)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "9932"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 32
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "378oLPG59TZx",
        "outputId": "e95f7619-c228-4d9a-aca4-7edd552fe240",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 289
        }
      },
      "source": [
        "for i in range(5):\n",
        "    print(unique_words[i], vector_list[i])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "preliminary [ 0.01543421  0.035304   -0.01397987 -0.02220014 -0.00381341  0.01415708\n",
            " -0.05078527  0.04488852 -0.05206664  0.00257026  0.02616317 -0.05171144\n",
            "  0.02606994  0.05129982 -0.05719861 -0.00063438]\n",
            "Externally [ 0.02713646  0.03406958 -0.01241592 -0.02222416  0.0056569   0.01001145\n",
            " -0.04102373  0.04381802 -0.04604217  0.01998601 -0.0028151  -0.02781722\n",
            "  0.02640389  0.04459162 -0.00765951  0.0191385 ]\n",
            "morn [ 0.00677163 -0.00337087  0.02717992 -0.01841573  0.04147504  0.00489579\n",
            " -0.05407092  0.06107125 -0.05222736  0.01113123  0.02121533 -0.04131475\n",
            " -0.02345313  0.02137208 -0.07194926 -0.00120875]\n",
            "harassed [-1.7969852e-02 -1.0123730e-02 -1.8228199e-02  2.0393420e-03\n",
            "  4.1150477e-02 -4.1259895e-03 -5.2599404e-02  9.7686965e-03\n",
            " -6.0656954e-02  2.7764378e-02  5.9136539e-05 -3.5123896e-02\n",
            "  7.3270034e-03  4.4308946e-02 -4.7052521e-02 -1.0095333e-04]\n",
            "short [ 0.04024915  0.25795814  0.18262678 -0.20742145  0.23201801 -0.02015576\n",
            " -0.83689225  0.53420275 -0.5880152   0.1744421   0.31806135 -0.28382793\n",
            "  0.04549685  0.72858775 -0.58712745  0.27769607]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CNeevc3jL6Al"
      },
      "source": [
        "# 2. Hypothesis\n",
        "scarlet is more similar to sin than to loyalty.\n",
        "Hester is more similar to sin than to innocence.\n",
        "Chillingworth is more similar to revenge than to freedom.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qFJgHbJZM8SP",
        "outputId": "c5ff90be-9dea-429d-b546-f5d533d3bd00",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 292
        }
      },
      "source": [
        "sim1 = model_b.similarity('scarlet', 'sin')\n",
        "sim2 = model_b.similarity('scarlet', 'loyalty')\n",
        "sim3 = model_b.similarity('Hester', 'sin')\n",
        "sim4 = model_b.similarity('Hester', 'innocence')\n",
        "sim5 = model_b.similarity('Chillingworth', 'revenge')\n",
        "sim6 = model_b.similarity('Chillingworth', 'freedom')\n",
        "print(sim1, sim2, sim3, sim4, sim5, sim6)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.9969873 0.7736934 0.99587274 0.4848858 0.98036754 0.97654855\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:1: DeprecationWarning: Call to deprecated `similarity` (Method will be removed in 4.0.0, use self.wv.similarity() instead).\n",
            "  \"\"\"Entry point for launching an IPython kernel.\n",
            "/usr/local/lib/python3.6/dist-packages/gensim/matutils.py:737: FutureWarning: Conversion of the second argument of issubdtype from `int` to `np.signedinteger` is deprecated. In future, it will be treated as `np.int64 == np.dtype(int).type`.\n",
            "  if np.issubdtype(vec.dtype, np.int):\n",
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:2: DeprecationWarning: Call to deprecated `similarity` (Method will be removed in 4.0.0, use self.wv.similarity() instead).\n",
            "  \n",
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:3: DeprecationWarning: Call to deprecated `similarity` (Method will be removed in 4.0.0, use self.wv.similarity() instead).\n",
            "  This is separate from the ipykernel package so we can avoid doing imports until\n",
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:4: DeprecationWarning: Call to deprecated `similarity` (Method will be removed in 4.0.0, use self.wv.similarity() instead).\n",
            "  after removing the cwd from sys.path.\n",
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:5: DeprecationWarning: Call to deprecated `similarity` (Method will be removed in 4.0.0, use self.wv.similarity() instead).\n",
            "  \"\"\"\n",
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:6: DeprecationWarning: Call to deprecated `similarity` (Method will be removed in 4.0.0, use self.wv.similarity() instead).\n",
            "  \n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "C40Aq1WSQgXn"
      },
      "source": [
        "# 3. Dimension Reduction"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WNrSjbrR9W2d",
        "outputId": "d7c04e40-518b-48a2-83ab-c584049b66f1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 646
        }
      },
      "source": [
        "from sklearn.manifold import TSNE\n",
        "# Lets dim reduce the 16 dimension vectors to 2dimensions to vizualise the dataset \n",
        "data_embed=TSNE(n_components=2, perplexity=50, verbose=2, method='barnes_hut').fit_transform(vector_list)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[t-SNE] Computing 151 nearest neighbors...\n",
            "[t-SNE] Indexed 9932 samples in 0.021s...\n",
            "[t-SNE] Computed neighbors for 9932 samples in 2.682s...\n",
            "[t-SNE] Computed conditional probabilities for sample 1000 / 9932\n",
            "[t-SNE] Computed conditional probabilities for sample 2000 / 9932\n",
            "[t-SNE] Computed conditional probabilities for sample 3000 / 9932\n",
            "[t-SNE] Computed conditional probabilities for sample 4000 / 9932\n",
            "[t-SNE] Computed conditional probabilities for sample 5000 / 9932\n",
            "[t-SNE] Computed conditional probabilities for sample 6000 / 9932\n",
            "[t-SNE] Computed conditional probabilities for sample 7000 / 9932\n",
            "[t-SNE] Computed conditional probabilities for sample 8000 / 9932\n",
            "[t-SNE] Computed conditional probabilities for sample 9000 / 9932\n",
            "[t-SNE] Computed conditional probabilities for sample 9932 / 9932\n",
            "[t-SNE] Mean sigma: 0.027449\n",
            "[t-SNE] Computed conditional probabilities in 1.277s\n",
            "[t-SNE] Iteration 50: error = 91.0126266, gradient norm = 0.0065804 (50 iterations in 7.220s)\n",
            "[t-SNE] Iteration 100: error = 83.1910400, gradient norm = 0.0021134 (50 iterations in 5.404s)\n",
            "[t-SNE] Iteration 150: error = 82.8065109, gradient norm = 0.0010953 (50 iterations in 4.563s)\n",
            "[t-SNE] Iteration 200: error = 82.6402435, gradient norm = 0.0010713 (50 iterations in 4.307s)\n",
            "[t-SNE] Iteration 250: error = 82.5170517, gradient norm = 0.0007203 (50 iterations in 4.252s)\n",
            "[t-SNE] KL divergence after 250 iterations with early exaggeration: 82.517052\n",
            "[t-SNE] Iteration 300: error = 3.3444552, gradient norm = 0.0012533 (50 iterations in 5.573s)\n",
            "[t-SNE] Iteration 350: error = 3.0700917, gradient norm = 0.0004633 (50 iterations in 6.296s)\n",
            "[t-SNE] Iteration 400: error = 2.9450750, gradient norm = 0.0002810 (50 iterations in 6.243s)\n",
            "[t-SNE] Iteration 450: error = 2.8688128, gradient norm = 0.0001907 (50 iterations in 5.996s)\n",
            "[t-SNE] Iteration 500: error = 2.8171332, gradient norm = 0.0001381 (50 iterations in 6.050s)\n",
            "[t-SNE] Iteration 550: error = 2.7795317, gradient norm = 0.0001075 (50 iterations in 6.033s)\n",
            "[t-SNE] Iteration 600: error = 2.7506528, gradient norm = 0.0000873 (50 iterations in 5.990s)\n",
            "[t-SNE] Iteration 650: error = 2.7281215, gradient norm = 0.0000753 (50 iterations in 5.895s)\n",
            "[t-SNE] Iteration 700: error = 2.7102485, gradient norm = 0.0000646 (50 iterations in 5.836s)\n",
            "[t-SNE] Iteration 750: error = 2.6958952, gradient norm = 0.0000578 (50 iterations in 5.769s)\n",
            "[t-SNE] Iteration 800: error = 2.6843696, gradient norm = 0.0000521 (50 iterations in 5.743s)\n",
            "[t-SNE] Iteration 850: error = 2.6751420, gradient norm = 0.0000514 (50 iterations in 5.693s)\n",
            "[t-SNE] Iteration 900: error = 2.6684301, gradient norm = 0.0000473 (50 iterations in 5.765s)\n",
            "[t-SNE] Iteration 950: error = 2.6631606, gradient norm = 0.0000446 (50 iterations in 5.634s)\n",
            "[t-SNE] Iteration 1000: error = 2.6587565, gradient norm = 0.0000417 (50 iterations in 5.712s)\n",
            "[t-SNE] KL divergence after 1000 iterations: 2.658756\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "smeJ3GXzQmJM",
        "outputId": "44c235d0-b0a5-4819-f21e-1277bea942e8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 816
        }
      },
      "source": [
        "data_embed_a=TSNE(n_components=2, perplexity=50, verbose=2, method='barnes_hut').fit_transform(model[random_keys])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[t-SNE] Computing 151 nearest neighbors...\n",
            "[t-SNE] Indexed 20000 samples in 0.785s...\n",
            "[t-SNE] Computed neighbors for 20000 samples in 361.961s...\n",
            "[t-SNE] Computed conditional probabilities for sample 1000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 2000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 3000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 4000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 5000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 6000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 7000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 8000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 9000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 10000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 11000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 12000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 13000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 14000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 15000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 16000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 17000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 18000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 19000 / 20000\n",
            "[t-SNE] Computed conditional probabilities for sample 20000 / 20000\n",
            "[t-SNE] Mean sigma: 0.260465\n",
            "[t-SNE] Computed conditional probabilities in 2.235s\n",
            "[t-SNE] Iteration 50: error = 94.1459045, gradient norm = 0.0457556 (50 iterations in 27.622s)\n",
            "[t-SNE] Iteration 100: error = 94.5396423, gradient norm = 0.0674006 (50 iterations in 27.387s)\n",
            "[t-SNE] Iteration 150: error = 94.2714920, gradient norm = 0.0811710 (50 iterations in 27.623s)\n",
            "[t-SNE] Iteration 200: error = 94.8409424, gradient norm = 0.0836515 (50 iterations in 29.079s)\n",
            "[t-SNE] Iteration 250: error = 94.7838593, gradient norm = 0.0867745 (50 iterations in 40.329s)\n",
            "[t-SNE] KL divergence after 250 iterations with early exaggeration: 94.783859\n",
            "[t-SNE] Iteration 300: error = 4.3984427, gradient norm = 0.0017002 (50 iterations in 32.077s)\n",
            "[t-SNE] Iteration 350: error = 3.9596279, gradient norm = 0.0018397 (50 iterations in 19.880s)\n",
            "[t-SNE] Iteration 400: error = 3.7955220, gradient norm = 0.0003318 (50 iterations in 15.616s)\n",
            "[t-SNE] Iteration 450: error = 3.7111843, gradient norm = 0.0002127 (50 iterations in 15.415s)\n",
            "[t-SNE] Iteration 500: error = 3.6588502, gradient norm = 0.0001475 (50 iterations in 15.400s)\n",
            "[t-SNE] Iteration 550: error = 3.6231854, gradient norm = 0.0001105 (50 iterations in 15.167s)\n",
            "[t-SNE] Iteration 600: error = 3.5971901, gradient norm = 0.0000884 (50 iterations in 15.239s)\n",
            "[t-SNE] Iteration 650: error = 3.5771818, gradient norm = 0.0000724 (50 iterations in 15.143s)\n",
            "[t-SNE] Iteration 700: error = 3.5618186, gradient norm = 0.0000613 (50 iterations in 15.016s)\n",
            "[t-SNE] Iteration 750: error = 3.5492592, gradient norm = 0.0000550 (50 iterations in 15.130s)\n",
            "[t-SNE] Iteration 800: error = 3.5387299, gradient norm = 0.0000468 (50 iterations in 15.022s)\n",
            "[t-SNE] Iteration 850: error = 3.5298715, gradient norm = 0.0000400 (50 iterations in 15.042s)\n",
            "[t-SNE] Iteration 900: error = 3.5222900, gradient norm = 0.0000366 (50 iterations in 15.103s)\n",
            "[t-SNE] Iteration 950: error = 3.5159435, gradient norm = 0.0000333 (50 iterations in 15.206s)\n",
            "[t-SNE] Iteration 1000: error = 3.5107412, gradient norm = 0.0000340 (50 iterations in 15.244s)\n",
            "[t-SNE] KL divergence after 1000 iterations: 3.510741\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6EEjFFMu0vGt"
      },
      "source": [
        "# 4. Visualization"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CA3AJtWfB-wC"
      },
      "source": [
        "%matplotlib inline\n",
        "from matplotlib import pyplot as plt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qfumg5VfCBfg",
        "outputId": "35364edb-f20d-4285-aafa-5a97845a402f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "x_axis= data_embed[:,0]\n",
        "y_axis=data_embed[:,1]\n",
        "\n",
        "plt.scatter(x_axis, y_axis, s=5)\n",
        "plt.show() ## The plots vary each time you run them "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXkAAAD4CAYAAAAJmJb0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3ujP5gMk1PWc"
      },
      "source": [
        "After dimension reduction, there is this obvious tail that extends from the main group. Thre must have been a group of words that were similar for there to be this shape since t-SNE keeps things close close together in the final visualization as well. Those words on the tail are probably very common words (he, the, is) that do not have clear correlations. This is not a very good representation because most likely, some of these words are only used once."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZMfazEYVErrW",
        "outputId": "ab8f5074-147b-434b-b818-8fe455ce500d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "x_axis= data_embed_a[:,0]\n",
        "y_axis=data_embed_a[:,1]\n",
        "\n",
        "plt.scatter(x_axis, y_axis, s=5)\n",
        "plt.show() "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXkAAAD4CAYAAAAJmJb0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0\ndHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOydd3xb5fX/38+9suQhecWJM20Hssl0\nnMloC0mAFuiiEFJGKRBKGe2XDihllzbQflt2gTRAaQsBSr8t8GvLCjOQZccZZBKIV4bjeG9Zus/v\nDw1r2ZZt2ZaU5/168SL36o5HlnTuec5zzucIKSUKhUKhiE+0oR6AQqFQKAYOZeQVCoUijlFGXqFQ\nKOIYZeQVCoUijlFGXqFQKOIY01APwJesrCyZl5c31MNQKBSKmKKoqOi4lHJ4qNciYuSFEOnAGmA6\nIIHvA/uAl4A8oAS4SEpZ29118vLyKCwsjMSQFAqF4oRBCFHa1WuRCtc8DLwhpZwCzAL2ALcC66SU\nE4F17m2FQqFQDCL9NvJCiDTgDOBpACmlXUpZB3wdeM592HPAN/p7L4VCoVD0jkh48uOBKuBZIUSx\nEGKNECIFyJZSHnEfcxTIDnWyEGKlEKJQCFFYVVUVgeEoFAqFwkMkjLwJyAeekFLOAZoJCM1Il3ZC\nSP0EKeVqKWWBlLJg+PCQ6wYKhUKh6CORMPIVQIWUcpN7+xVcRr9SCDEKwP3/YxG4l0KhUCh6Qb+N\nvJTyKFAuhJjs3nUWsBt4DbjCve8K4NX+3kuhUCgUvSNSefI3As8LIczAF8CVuB4gLwshrgJKgYsi\ndC+FImIYhqS62U6W1YwQYsDPUygGm4gYeSnlNqAgxEtnReL6ihObgTKohiG55E8bKSqtZW5uBmuv\nWYim9Xz9vp6nUAwFStZAEdV4DOqiVetYvnojhhG5/gfVzXaKSmtxGJKi0lqqm+0Dep5CMRQoI6+I\nagbSoGZZzczNzcCkCebmZpBlNQ/oeQrFUBBV2jUKRSAeg+oJjUTSoAohWHvNwl6Hgvp6nkIxFCgj\nr4hqBtqgappguM0yaOcNBmpRWOGLMvKKqCeaDWq0oRaFFYGomLxCEUeoRWFFIMrIKxRxhFoUVgSi\nwjUKRRyhFoUVgSgjr1DEGWoNQ+GLCtcoFApFHKOMvEIRYQxDUtXYjkthW6EYWlS4RqGIIEORwqjy\n4hXdoYy8QhFBQqUwDmR8XOXFK3pChWsUMUm0hkQGO4VR5cUrekJ58oqYI5q918FOYRxIbR9FfKCM\nvCLmGOyQSG8ZzBRGlRev6AkVrlHEHKqq0x/PQ0UZeEUolCeviDmU96pQhI/y5BUxyUB5rw6Hwd4j\nDRiGEdHrKhRDhfLkFQo3DofBnPveprHNgS3RRPHtSzGZlB+kiG3UN1ihcHOgqonGNgcAjW0ODlQ1\nDfGIFIr+o4y8QuFmUrYVW6JrcmtLNDEp29rjOSq8o4h2VLhGoXCjaRrFty/lQFUTk7KtaFr3PpAK\n7yhiAfWNVCh8MJk0poxK7dHAgwrvKGIDZeQVij7Sl/COQjHYRCxcI4TQgULgkJTyPCHEeOBFYBhQ\nBFwmpVTCGoq4obfhHYViKIjkt/JHwB6f7QeAB6WUE4Ba4KoI3ksRx8TSYmZvwjsKxVAQkW+mEGIs\n8DVgjXtbAGcCr7gPeQ74RiTupYhNwjXcnsXMcx7+iFn3vo3DET2GPlqVLxWK7ohUuOYh4OeAzb09\nDKiTUjrc2xXAmFAnCiFWAisBcnJyIjQcRTTRmyyUUIuZU0alDuZwQxLNypceQjUPUQ1FFP325IUQ\n5wHHpJRFfTlfSrlaSlkgpSwYPnx4f4ejiEJ6k4Xit5hpMZGRnBAxz7k/YaBA5cuqpvaQXv1Ae/td\nXd8wJMtXb2ThqnVcvHojDodBZUMby/+0kUWr1rF89UYMQ81ATkQi4cmfClwghPgqkAikAg8D6UII\nk9ubHwscisC9FDGIx3B7PPnuslA8i5n7jzVy12u7OPWB9yLiOTscBnN+9TaN7Q5SzBovX7uYydlW\nalsdYXm5vrrt+Tnp3Li2mK0BXv1AefuGIalsbONYbSs//cd2Pj/ewtzcDF5auQjDkOyrbKSmuY3N\nJTUAbD5Yw/mPfsieymbvNQqjUJJZMTj028hLKX8B/AJACPFl4KdSyu8KIf4OXIgrw+YK4NX+3ksR\nm/Q2C8Vk0hhuS6S4rK7XmvFdhSf2H2uksd01m2i2G3zt0fUIQAJzc9L4+w9O7dYge5Qvq5raqWlq\n57zHPsYZMLaB0Lk3DMnFT33CltI6v/1bSmr54LNKbnxhG03tzqDzfA08gMUkSE/U+zUWRWwykCkB\ntwA3CyEO4IrRPz2A91IMMXa7kw0HjuN0Bhsc6D4LJVQIwuM56wJmjEmlqqG1xzCLx5MOFZ5IsQTf\n1/NqUVk9r24v73Lsvty0tpjzHl1PsllHD9Czz7Kayc9JRxeQn5Perc59S0sHf9t4kA/3HuFoXUuX\n4Z3qZnuQgfdw5bNFIQ18yPvZDWb96h2aW+y8v6+SV4vK+Gjv0bDesyK2iaisgZTyfeB997+/AOZH\n8vqKgcPhMPqc7223O5l61xs4JegC9txzDmZzeF5jVyEOIQR/vXI+33ryE4rL6/nqox9jtehs/eVS\n6ts7Qyy+nnuouLkmBKlmna/874fdjuN/XtrJzS/tZN+9/mMPdX2nhJZ2B6/feBpZ1k5P3WWnBQjX\nf1K6/hlIS0sH0+59y2/frFHJ/PPGLwfNJrKsZqaNsrL7SP+raVvsTk659+2g/fvvDf/zUsQeKrlX\n0e+0xaIyl+EDcErXdrh01YjaMCQXr9nIp4cbvMc2tTv51pOfsHDVOi56agNH61pZvnqD13PPTE7w\netJzxqVx49piFq1axwWPryecNUcJbCmr8W4HzgzSE03MHJuGLmBubgZ3v76bxfe/y8VPbaSyoY3j\nTe2uv0UPTbVf33kkaN/2Iy18XlkftF8IwWvXn0ZywsD9VIvKalV6aByjjLyi3xos88e7wirg8uTn\nj88I+9yuQhzVzXZ2VPgbvaQEjU8PN+A0JFtKaln8wLtsLul8QBxvsgMCCdS2dD489h1rDr5xF0gp\nvWEh3wdQYUkNF/1pI9sr6pk5Lp1HL5nDVvdrm0tqWLxqHdc/v5XkBJdHnGzWyUxOCHmPb8weGXL/\nJV1kwJhMOjvuOpt/Xr+IvIzIL5zOzUnrMsyliH2UkVf0W4NF13X23HMOa69ewP77zkHX9W49Q99U\nxs6XBVJK9hxpwOFwYrc7GD8sye+81g7/GYavLZozLh3DMCgsqcGQcKCqBache/UF1wRc+vQW72zG\nt5fszLHp7Kiox2lIdlbU45SSqaNs6O7wilPC1vI6mj2Lu20OPjvWFPL9N9hDG9Hjrc4uvX+TSWNs\negrlde29eEc9I4DlazZTGGI2pYgPRDRNzwoKCmRhYeFQD+OEpD8x+UA8YY7C0lpmjEnlp8smseik\nYei6HlQY9eZNp7P4t+/5na8Jwgqv+DI+PYFjzU6aAx4EZl1gd/Z8sSdWzOG6F4q92/++8TR0TTBh\neArVzR3c8EKRdwG0ICedfcdcs58Ui87UbCvbKhrIz80A6TKUyRYTLXYnBSFSKaWUXPjH9RSVN/iN\nYW5OGq9cd2qX6ZxSSi5+agObS8IPh4WDLmDmuHR2VtQzNzeDF1cuVIVTMYYQokhKWRDqNaUnrwBc\nRv5gVTMThif328hXN9spLHXFpreV13Pp01u8C7JfVDf7h4aONwad35dowcG6jpD7uzPwnhTKebnp\nLDsl25vLb7WYuPipDTTZndgSTbz14zMoLneFjnQBNy+bxIo1mwFobndy7zdmMNyWSJbVjJSw92gD\n54dIsfTeVwj+ft1pVNQ288H+StISE5iXN4yR6cndGlchBC+uXMSLm0q47dXdvf8jdUFBXiYvXL2A\nmpYOVRkbh6hwTZwSGC7pLsWxrc3BlLvf5LoXtjLh9jdpcxvhcEMugWRZzcwcm+a3z7Mge9KwZL/9\nq/6zt8/vsb943pXQNKQU/PfG03jsktnc/41pNNldf6fGNgd1LXZv2KYgL5MF4zP9wltTRtr8morf\n/dounO4nVX5ORshUSk0T5Ayzctmik7lgTg6jMlLCMq6aJrho3rgIvHsXCcBfvjdPGfg4RnnycUhg\nWuJzV8zjlHve9KY4br99KdsP1zNhRApZ1kT+sa3C7/z39ldx9vSRXVZvOhwGs3/1Fk3tThJNgndv\n/hKjMpKREm+64SvXLuJbf1zP9kMuT10TUJCbTmFAzveeo4PTaCM33UxpXehYc1FJDd95agPF5cH5\n6LZEE5NH2lh7zUKqm+2kmnU2H6yl8BdncrCmNSi85VmsBddM4Z4LpkX8vRiGwGwS2B39D7V2AF97\n9AM+P95GQZ6rijbaNHkU/UMZ+TggsMqzsqGNLe4FyKLSWt7df8wvxXH6rzpzpZMStKAFzZOzLK50\nwBDVm4Yh2XDwuLcIp80hWfzb95ky0kqHw6CkppWC3Ayev2oBCXqn8Us0aaxYs4mi0jp0AWGEySNK\nVwZeANNH20Ia+CdW5HP29GxAeA38tLvf9KsHCAxtZSYnkGxxhX00Aec99nHIuHxfcTgMzn3o/YgY\neA8HjrcBriraXYfrmD4mXXn0cYQy8jGOr9een5PBgxfP4uyHPvTGtSeOSOZPHx7o8vxAAw+w7JEN\nzBxtZc64dIrL67xVnR4RLI9Gii97fTzywtJaDlQ1sbWsMwWypcOgqLQOAzrjJFHAw8tn8eOXtod8\nbcnU4YDw/n0njLAG1QMsmpDlt2hd09JBizvU45R4F2IjJXHwndUb+Lymrc/XsOiC9m6esOc/9gnz\n8zJ4UXn0cYOKycc4vrncm0tqOPWB9/xK3fccbaa4onNxc5Q1vOf6jsNNtNgdrL/lK95si+pme1iF\nTicNS2FStpW5uenefTaLifyc9G7OGnwEcNOL27tc6P3GHz/howPHKHTn4n92rMn7g9GEKxvmUG0L\ns+59k3Me/oipt/+X9/cd5qTMRACSTAKBK/8/Mzmh38VGoWoHPEwfbeOf1y2gJ3ma7gy8h6LSOpVG\nGUcoIx/jZCYnMGNsmjdfuyeONDl6PsjNriONXPH0JpxOl7efatYx6z3f5z83LUbTNNZes4jZY1MR\nwJSRVh5eMSeqvnA9mbtdRxq54plCnG7DbDFpzBydArgygPLvW8epD7xHs93192k34Gev7Oaz460A\ntDokErA7nHzj8Y+Y/+t3uOjJT/pcbJRlNVOQm4GuCWaMScVTBJti0Vn7/fnsP9pC0W1nMTk7xe+8\nicOTQ1yta+bmBuvuqIrY2EXlyccwnlDN5oM1mHUIU6uq1yQnaLz5P2dwzkMf0Wzv/ibPXj6HRLOZ\n+eMzqGlxsGjVOhyGdOVij00PGfs+0Xh4+UwW5A0jOy2p17Fv3/WXjg6DorJapmWnMOvX74Y8ftpI\nK3srm8JKS/Wsz8zLTeexFXMZkWrx6gNFe8OUE53u8uSVkY9hqhrbWbhqnTddLxII4LrTcvjj+rIe\njz1pWBJfVLd6tzVc3rHEtTC5++6zufzPWygqrWVqdgo7IyCy1Rd0INSjqSAnnZuXTWJMmoUv/f6j\nwR4W+WNtvPLD0wFCyiN3h2FILnpyA1vLa9GArtZhZ41JxWzSulSy7ApdE94F4+pmu/dhbdIEG35x\nVq/WF1R3qoFHFUPFKcm6YFSqhYq6vi/EBSIhLAMP8EV1K8lmnS23nsmOww04pcGlT28BOsv8116z\nkEN1LZz+2/cjNsbeEsrAp5h1hIDLntkyZCGIrRWNHGlo5eaXtvfKS3Y4DD7cW0mhe32kOzm5nYcb\neP+nX+KM330Q9FqyWWfy8CSKD7kevrq70liCXyGXb8MUX2nlcFCzgKFHGfkYpaWlI6Rs7KCPw+7k\nQFUjiyZk4XQ6vemRAqhpbsMwnOwoD71YOJQ0252d2T5doAm474JTeG5DSa9EznrDB3sr2XKwBgN6\nzMLxdIha9ocPvQ1QeqIgN4PtFaG9+Ba7k6vOmMANa7cBrs/thavn89C6A96uVx7v2+PR99YbH4hG\nKoreoYx8jOFZAHutOHq6KX79iY18dMuXGZuezJ57zuGTkuN875lCrl+7HdjO3CjLqvEQysAn6XDK\nmDSKKxpINAlue3XXgI7htn/tRsMV6urOS3alr25gS0lt2Bmof72ygMmj0qhqaOnymD99+AU2i4nG\ndgc2i4mTh1tZG0LiQNOEn3EONwTTn1mAIjKomHwM4WoFt4EtpZEVqArF6FQzhxt6l0bnya/+z6eH\nueGFbd79fREcGyo0AZ/ceibHm9o579GPB+2+a69ewMKTh3VpMCsb2ljwm3W9umaKWae1w8mMMal8\nfqyZJrszqPhNE/DaDaciENz92qcUl9eHrHD2Fa/zr81I59FL8r2LtKFQMfmBp7uYfDRltCl6oKqp\nPcjA//CMPF6/fiGR/unc/+3pIfdrAuaOS+Xh7wS/vrmkliP1rdzyyg6//RbT4H7NEk19/2sY0tXi\nTxtkYzQ3p+sqU8OQ1DT1XmK42e7EkLC9wrVeApCXmcj0kZ0ploaErz/+CT96sZgtpa6euoWlteyv\nbERKGbKhjH9thkvXf/nqjTgcBlWN7Tid/v8XAj9tH8XgosI1MUSon8gfPyxhS2kdM8eksv1QQ4gj\nek/+uFTyskJryieZdV5cuRghBM9uKGfHoXo/L/3vW8q9eeMQnIEz0EzNTglqYg2Qm5lEggYHjvc8\nlqKyOoalmEkyCVq7SFvxKFhGigPHm5k2Oi1ov1e2uaQGm1u+2NmH2Xdrh+ucPZXBoRunITlQ1fk3\nSzJpfO2Rj5ibm8GPlkwMaigzcYSVGWPT2FZW512k3XLQpf+z41A9yWad5jYHKYldyy0rBg/lyccQ\nw20WCnKD49tbSusiZuAB6lsdnPG70D1Rm9udfFbVxEWrN7Ctoj4oDPPQu50SCppgUA08gJShl1JL\na1oxhVHIBTAtO4WaZjtbf3lWlzOkSEefMpMTqGxo41hDm7s7lWvtxaMh5JTQYnfw7PdDzsgjSpPd\niVO6ZmaXP73Zu18DrAkaFz75CcVl/ou5BlBcXofTkDS2OTBwPRScATMDUIVVg42KyUc5gfFMw5BU\nVDfzyefHebmo3D0VH7zxJCVovHTNfC7440bvvumjU6lpbuNwfWcM/8avjOfR9w4O3sB6ybNX5JOR\nmMC3Vm/qcr3AZjHx/25czPef3czn1a40VYtJ0B5BcTAAq1lnykgbhW7DOT/P1T5xa1kd+bkZOB1O\nisrrOWWUld9+eyZfe+yTiN5/oPGkZs7JSeelaxZy6TObVUplhFHFUDFKqBxjwE+Q7JFLZnPVc1vY\ndTi4+cZgoNF9nnas45JIzuThS2YjAMNwcuoDH0TUkxeA8Fmc9tg8QwYvWqeYNaR0Cb7FIlNGWvns\nWDNOQ6Jrgo9//hXqWjsi0pHsREYtvMYooXKM/faV1VLX0sGfLps7ZGOMZlMzMSup54N6wHCrTXY4\nnFy6ZhOn/fYDZo1N7fMPJ8UcrCA2OTuF/HGd8XhDuv7TNcFJWf46NM12g5euXUhyQs8jiEb/eO/R\nJm+FdlKCzll/eJ9zHv6Imfe8xdG6VhXCGQCUkY9ifBtJe3KMPft0TZBs1vnaIx/xoxe3YQkz3nwi\n8dnxVpJ8Mm0E8PHPTmfmGJvfcaeM6tyekJXMnruW8O8bT/MqOpqk5IzffcCBqmZvtsrUUf7XCDeD\nwWk4vWMBl6e+t7KZfZXN3H3+FL9jf/P16bz+w0V++6wWnRG2RLbfuYz/3nSaN7QDLo0hz0PEatGZ\nl5cRtnDdUNDU7vAu0je1Ozn1gfdYvnpjnwXcFKHpd7hGCDEO+AuQjWs9arWU8mEhRCbwEpAHlAAX\nSSm7TfBW4ZpgQuUYG4Zkf2UjX3vkI28DC4hcI45QjUTiBatFZ3K2jSJ3/HvmaBs7AkJd8/MyeODC\nGXzlf0MvPg8UyQmaXxhm791LmXf/ezS2OTBr8PI1C/j1G/soLq/35qdnJidw0eoNbC+vZ+bYNB6/\nZA4N7Q5y05N4/8BxZoyxcs1ftrI3RMbRUBMqQ0nXBBt7qY2jGHjtGgfwEynlViGEDSgSQrwNfA9Y\nJ6W8XwhxK3ArcEsE7ndCEVhp6Nk3eaSNgrxMb0EKuLIh+sucnHRevmYh+ysb+dkr29ndz/Z8M0al\nsPNI9BiYpnan18ADOEJ4jZtLavnHlvD0eyKJr4F/7nv5bC2v96Yv2g1odboWYz2ZL4vvf5eZ49LY\ncagBA9hWUc+pv3ufgpx07yKuh2gsSPMdjiddNTFBI82iU9XY7pdsUNXUjkDl2/eFiC+8CiFeBR5z\n//dlKeURIcQo4H0p5eTuzlWefHjY7U6KymopyE2nrs1JltWM0ynZd6yR2/6x3dtXtTdoAmaPS+eV\nHyzyVjVWNbVjOA0OVTdz4ZrNPV8kBKeMtqEh2Xl4cBUoJ2clceOSidzwon9hVqIJ2sKX1I8aBPCP\n6xZw6dOF3s5THiZnW9lXGd7fd2yamYr66G4IkmLRabM7KcjL5PmrFrBizUavAzM/L5MXV6qMnEAG\nLbtGCJEHfAhMB8qklOnu/QKo9WwHnLMSWAmQk5Mzt7S0NGLjiUfsdidT73rDr8+oyaR5C2b6GrLx\nePB1bQ4ykxNYsWYTmw/WcFKmhdLa9i6lbGOJJJNg2x1LmPubd2lqd0aldxsJuioIizU04D8/Ot0b\nlgQVzumKQZEaFkJYgX8AP5ZSNvhOqaSUUggR8uckpVwNrAaXJx+p8cQrRWW1QX1GJ2Tb+mXgAXaU\n13HRnzays6KeGWPTvMUun9f0vpw+Wml1SGranN44wVAZ+MDYe18IFc/WgMQEzWvgI12VOxjomvBm\n3wgBJw1LZm5uhteTVyJnvSciRl4IkYDLwD8vpfw/9+5KIcQon3DNsUjc60Rn/vgMr5yvLlzbIEi2\nmGhsc3Rbit8VmoBZ49LZXlGP05DsqKjnpGGJfFEdGZ16TcBL187nO0/2LeQTKZISBDVN7TT10N1q\noEk1C1o6+neNwE84KUHD7pB+D49YM/AA2Skahxs7G6EfON7MiysXqZh8P+h3CqU7FPM0sEdK+Qef\nl14DrnD/+wrg1f7eSwG6rrPnnnNYe/UC9t93DrquU9PS4Y3TtvUhrvL4Jfm88oNFFHjSNXMysCRE\nTtbIkHD500O/1tLaIbn1H9sZ6nDu0Wb/h0xuRv9DD3an7JOmTbThMfAeMpISqG62M8JmYURqojLw\nfSASKZSnAR8BO+msjbkN2AS8DOQApbhSKGu6u5ZaeO0bUkqWr3ZVwc4Zl0ZhaV3YXpwA9t6zjIM1\nrUwYnkJtq4O2tg5O/31wJ6F4wTMTihYevGg6//Pyp/26RjyuL0wbZcVqSWBrWZ2SQOiBAY3JSynX\n03Vx3Vn9vb6iezxa389fNZ/aVgdZVjNH6lo59YH3ejT0E7KSef36UylY9S6NbQ5siSaKblvC8ifX\nR3ycJg0chqvl3JQRKWytiJygWm8xm6KnDsCiw0/+3j8DD/Fn4JMSBE9fXsDpv/tAdZXqJ6riNYbx\n1fqec987ZCQlIIQgOzURa6Lr+Z3iUwUZyBfVLRRX1PtJyW4qrfaKcUV0rIbLE2ixOymOsIEfY+ud\nrxItBh6gw9k3Az1rTCq6Jryhp3jzb9s6JD96aTv5Of4V34reo/TkY5gDVU1BWt9TRqW6YvTuHqBt\nDoP1t55JXbOdO1/bRWFJLYnuilZDwkPrPiPFrNFsN0g2a2QmJQzYeGXA/yPFocYYTHx309fHzf5j\njfzzB4u44I8uRco4c+SRuAq/Pr71TDQhVFepfqA8+RhmUrYVm9tjtyWamJTtavSRZTVTkJeJSRMU\n5GUyMjWRqaPTWHv1QubkpNPm48kWldYycYRLh6XFbnDX67u79PyjFUsMfItPyui9F2oOeF9ZyZ0P\n4NYOycV/2tDt+Rr4afcAnDw8udfjGCryc9IZYbOojJp+oqSGYxxPTN6zaBpYCu40DOpbOpg80kZ1\ncwcLV63z5iHrmmDqKBt7jjR692nCpd64r2pwm30owqO/0s4f/eR0Tv/9R5EazoCy4ZavYDLpft9p\n1Ss2NINSDKUYGkwmjUnZtqDGyllWMzc8X8SWUldRU7JZZ8utXyExQaO53YnVrDNhRAo7KhpIsZho\ndId3ZoyOTBvBJJOgzSEZl5FIWW3kY/wnKv1dTSiNoc/iB38r4tPDDczNzeBv31/A8jUb2V5eR0Fe\npre3QlVjOxKJNCR1rS5nBoR6GPigPPkYxtNGraa5nfMe+9jPQ5/p04OzK3xTCT1VmFaLiYnDkyiu\nGJomJIqBJ8Wi09w+tAVhvSXHx1kQwB3nT+KpdQeobPF/7CWZBJNHpbLzUMMJ1VtWdYaKMzzG/YYX\ntrKl1FXubTXrNNudXqOua4Lpo2xdeuWagDljUykqD3794YtmUtvaxjMfHqSsPnRpZqLbU/cwaXgy\nn1W1xN0CYLwyY0wqu480gpRRVTPgIRKSDCeSzo0K18QRnpaAhaW1Xs8dXM2XJw1PZn9VCwD549Kx\nO7rOOjEk3HDWBK7889ag1370cqdy4wc/+xJVDS1c+NQWv2PaHNJPQsFz38EgFjVZoolhSYJXVi6i\nsLyW37+xh60Vjd5iqmgpqorEEKSUHK1vYVhKwgndWlAZ+SjFd5FJSrz/rm62Bxl4D76GdltZLR09\n/FKufS7YwAeys7yOeeOHMX1kCp8ebfZb+OutRk6kiAIbFDNMHZ7Inir/OPxF+WOZc9/bfjo3nq+T\nIWFqdjJ7KgfvoT1QGBLOf+wTks0a225fitl8Ypq7E/NdRwGerJhQDYx9G3jn52QA0lva/cLVC5iS\nbWXXke5j5j0ZeAB7GMfc8OJ2/7H1fIoiirhx6RQmZyVx1iOd6ZZPfFze7TkarlBHgpC0xVboPiQt\ndoPZ973DjjuXYTKdeB69MvJDgKdS1SMlUHz7Ur8vX2Czbk/ctKi0luPN9gHJGDg5K5nPj8e+9xar\n6MBA2NMfvrCt19Wwu9xefF/HE43htBa701sseKJx4j3WooBQlaq+BDbw9v23APYcjXzmizLwQ8tA\nOsyDbXCjzcB7OGlY7BSCRTxMCkEAACAASURBVBLlyfeCSBVjeCpVPZ68p1LVgxCCtdcs7IzDN7Ry\n7//7lEkjrNjtdubmZlB4sEaFThSKXvBFdcsJ6ckrIx8mvnHy/sqeaprGllvP4r39VSydNjzkyr+n\ngXdNQysFq95z763if9cd5OTMBEyaq7mzQnGiMHdcKj9eNpnLnt7S88EB2CzBztSJgjLyYeIXJ++n\n7KnDYTDv/nVdxuR9+d0be4P2fV7Tz7ZC3WDRIcbqZBSDwOhUC4cbhqYVZF6GhZLadnSTyauzFC6v\n33gqJiGYPNJ2wqZRnpjvug8Exsn7I3vaU0weXA27X99ZxktbD/f5Pn1BGXhFIJpgyAw8QEmt696b\nD9Zw4Fgjs8bYEEBSDy7q7HFpnDIqlckjU6lu7iCaCj8HE+XJh0lgnHwgY/J2u5NJd77R3yErFBEh\nGoqjPFzqDtXMHpvGi1cv4MLVG9l1uAFrookWu5O5ORk8dPEsrn+hmB2H6rnoyU9oanewt7KZvDQT\nM8dm8OUpWVwwJweT6cQwf0rWYIjoLk9+w4HjXLJm0xCNTKGIfjRgVk46OyrqmTk2jZevWUhdm0uF\ntbK+jcX3v9tjYsKuO5eQkhwfkgfdyRqocM0QYTJpTBmVGjJOOH98xhCMSKHoGatZZ864NHRNDGk3\nKgMoLqvDaUh2VtRT1+ZguM2C0yk58/fvhZV5dsq977Dpi0r+s+MQdnuHS9EyipzeSHFizFdiDF3X\n2XnHEmb86p2hHopC4ccomwldiJCyGkOBHrBGdqCqiZZwyr3dXLy6M3KgCZiXl8nzVy2gtrUjbqSK\nlZGPUj7tQbZAoRgoPNksofisuh2qh24R1pdTRtt49nvz/TpHTcq2kmwStPRBV8mQUFhSw3dWb2Bn\nRX2/U6WjBRWuiVLmj8+gt9+thAH6NGP7K67oDbPG2Ciriw4j3h3JCRqvXX8qI1ITOzuhNbYjJUwc\n2beCJ03ArHGuOL9vqnSsozz5IaatzcE7+yoZn5XM5OxUbws/XdfZettZLHnoA443h9eoumOAiqOi\nY2KuGGhSzDo3nHUyK/+yrcdjLRq0D0IxXqJOFyJpEsOA9vYO/rnjEH8vLGfnoQZmjElle0V4nc00\nXN/tk7KS+fcNp9LYIRmWksAlf9rkLXrsT6p0tKCya4aQtjYHU+5+07utuZWd5o3P5NnLCph271sD\ncl+zCE+BUqEIxczRVh5dns+X/vDhgN/LBHTl4gxLNlHdEp4D1O09NMEGn+YisdhLdkiza4QQ5wgh\n9gkhDgghbh2Ie3RO1WLLcr23v8pv25CurIGi0lr+tePQgNzToisDr+gfT1xawIYvagblXt2Z8FAG\nftqoYOmCFLPe5TV0QZDH7pEUiRUD3xMDGq4RQujA48BSoALYIoR4TUq5O1L3iKSmzGCzdNpwv22P\nJz9nXDoW08BY4nAqWpWnr+iKsakJnP3wRzRFYWm0RRe8fNV8Ln56M7uOdFaRt3Y4yXfn1M8aY6O+\nzcmBqmamj05lzeVzyU5LGnSDPpizhQEN1wghFgF3SynPdm//AkBKuSrU8X0J11Q1trNo1Tochgya\ndsUCgTH5qiY7S/7wAc32rn9EJ2clkWI2seOwysBRDBwJwB0XTObO1/YN9VB6TbJZ45TRaRT7NNs5\n1thOXYud8ZnJzP3NOprtzh61owaCgXBMh7LH6xjAtw1NBbAgkjfwaMrE6kJJYqKJ82aN8W7Xt3Z0\na+ABPj/eSvJApdIoFG46oN8G/vrTc3n8o9LIDKgXtNgN7rngFIbbEsmymunoMFjy4Ac0tztJNuu0\nuH9jjW0O9h9rZNrotEEbWyTFDsNhyLNrhBArgZUAOTk5fTk/YpoyQ4VhSKqa2hG4OjSF01mnZaBS\naXzITTdTWhf7KWSKgaHrzJdONpXW9UrZNMWs9+jkhIPNYmKKW3nSMCTffuoTmt2DaAm4/rCUwXUM\nB9sxHWgjfwgY57M91r3Pi5RyNbAaXOGavtzEs1ASixiGZPnqDWwuqQVc8fhoCYcrA68IxDfbpc0J\n44clcbC61fu6wFWQtK/SFRMvLKvn1esWsHzNFlrDcEwiYeAF8M7NZ3glQ6qb7ew61JlWmWLWmZKd\nQnF5A3Nz0xmRmtjve/ZqfIPsmA60kd8CTBRCjMdl3JcDKwb4njGFZ+rmYceheiZmJfHZ8dZuzlIo\nhobAfBZfAw8uA3ugqtlv39ef2ET+uDTqWzsGpc3knBx/w51lNVOQl+F1pKaNsrH2mkXUtnaQmZzA\n8abBjwIMpmM6oIFdKaUDuAF4E9gDvCyl3DWQ94xWukrz9EzdPOTnpJOY0HXKl0IRzRjAxBHBaYxb\ny+v54ngLKRbXd3vSiO77rVq7SXvsjinZKbzyg0V+VbAAj16Sj+5e3Cwur6e2tYNhKWZWrNnEolXr\nWL56I0aU6PFEmgGPyUsp/wP8Z6DvE810t5ouhODFlYu8MXlDShauendoB6w4obhq0Tie3lDe7TEW\nk0a7ozPckpthJiMliW0V9YAr/deQrlj4q9ct4ltPbmBXgP6SBJrbnfz3ptMYk2ph5n3rvK8FrkM1\n2Z3ea/YGW2ICIIJ+cy9cvYCCgDj48abBXQAdKoZ84TVSRHOVWk+r6ZomyHZPL4/WR1+YRtcE00en\nsvNQPXnDEvnieNtQD0kRQXoy8ICfgQcorbVTWmv3Lqqa3T2Hp4y0ctmzW9h7tJHZY1P50ZIJIOHK\n57Z6z73j1U85e9pIv+uFsuUWkxZWHH/S8GQ+q2pB4vLSPXozvr+5mpaOoDh4rGfmhUtcGPloL4gK\n98tkGJKbXuxZN2QwEUD+uHS2uNcNQhl4HYi+0hjFYODJmvFk2WwtrwcpcUrYcaiBG17YHrSYurW0\njocvmsmv/+vfvzgpQdDqIxPc1mGElWn24MWzufffe4N+X4G/OSH84+DRlJk3kE5qXBj5wc477S3h\nfpkCF2GHgsAfVUaSie8tzvUa+VDIcH6JihOCqSNtJCVobC2rY2K2lb1Hg/sXzxmXyuhMK9t/eSaz\nfu0KTeoCtt2+lC+qW7jrtV1sdRcx/eE7Mzn9t+97m4AIYOYYG58fb6HJnfM+ZVRqyN9XOL+5aMjM\nG2gnNS6MfLROuwJb/PX0Zcqympk51lWlN1QE2uqaVgdtHd2nUiaatEHJ21dEBl2Ac4Aeyp8ebmBe\nbgYzxqSx41B9yLi6gYaUkGZLYv+951BUVsv88Rnous7U0Wm8uHKR1ziDS7CvqLSW/JwMHlsxh+E2\nCx0dBt9+6hN2H2lkxZrNrL1mYdDvKxoMeDgMtJMaNyqU0RaTdzgM5tz3trdZd7il006nwZd/9x7l\ndYMf9xa4ysGb7aENdoIu6IiQdVDOf/wS+Nn+9fvzuP+/e70Lsbom2HDrmWHlpzscBvuPNTIsxezV\njofYlzPxRUrJ8tWdnvyLKxf22oadED1eo0057kBVE41trqzixjYHB6qCp62h0HWNtdfMH8ihdYkE\nThqWBEBSiDlepAy8516K+CTws508MpXXbjiNOePSAXAakpV/2cLh2haONbThdBoh04s9jtJXH1nP\nkgc/xOnz/fPM3k0B7f9iEU84d8MvzuqTge+JuAjXRCOTsq3YEk1eT35SdnDucFdkp4YnbTAQ7Dzi\nKmRpdcBkn8pFhaIvTB9t8zpfT142l0W/WYcBbKtoYPED7wGuCtQ2h8HEEVZev34xCQkusxTKUZoy\nytX1KZoWTSPBQIaW4saTjzY0TaP49qW88aPT2X7nUm+JdTh8frw5Kjzd+795SkSukxCRqyhijSST\n4P9+sIhjje0ca2hjuHvNKZBmuxOnIdl7tJHJd7yJ3Z2N43GUgJCOUrTN3qOVuInJxxOGYTDznrdp\nau8sIveIQU0fmcyy6aP4wzufD/g4BqvFmyI+EUCKRfdqz8/Py+T286ZwwWOfdHve2qsXsGhCFhCc\nvOAh2tbgesNAjP2EiMnHA54ybFcVrL8isycPeU9lC3/bMDjSrcrAxy/j0gd+kVKCX3ORorJa9DCM\n2vzxnTIfJpPmDtEIb9zek3IYi3IEQzH2uDPysdoKMPDDn5Jt805VfZk5Np1jAY29A3832SlqqUXR\nPeV17YN+z7m5GWT2IOu7564z0XV/3Rrf38bFT21g79EGCktqcBiSwpIa9lc2xszvPVS65EATV0Y+\nlp/wgR9+bauD4tuX8vzV8/HYcF3Ak5fmU5Cb7j1PAK/+wN/rX31Zvt92YgT1zu79+hRy0zuj7NNH\npzJrjK3L40fb1APnROev3y9g821n8dLKhd0et/by6SQlJQXtr2psp9D929hcUsv5j32MxS3iZzEJ\nznn4Iy7843oq69ui3tgPRVZQXP0Co73ytTt8C7ryc9KRUqLrgoXjhzEnJ53t5XUU5GUyIjWRl69d\nzJH6Vkqrm1kwPhNN05g7LpWi8gZmjrYybUwGyQmdBUptTtCFwBmBH8Cdr7pK0QUwfbSVnYddOt0z\nR9tCtiM83NhdK2ZFrNCfbK/rnt/GtjuWIoSgrqXD7zUNl3Jlog4LpnS2nvDErTOTE7hx7VacPg6b\n05C02J0IoMUtg1BU3sDCVesoyMvgsUvyGZEanQuyQ5EVFFcLr5EoKhhKPB2iblxbzFZ3hR9Iisrq\nmDk2jVeuXYSuh558ORwG31m9gR0V9cwck0ZxeWfVbHKCxkkjrHzq0zgh0mjC1Ui51RE93ydFZJg9\nOgWTKYHCMCuxzcDN50zi/jf2e/e98aPTmTIqFcMwmHXP2zS2O7BZTBT98iy+qG7xW1j1LfOfMTaN\nHeV1OKXrQZOY0LNoma4JCqJQw2ogOWEWXge6qGCg0TSBJgRFnnhjaQ1FpbU4DcnOinpqWjqC1hzq\nGtv49f/bxdG6RnZW1OM0JDsO1ftdt6XD4KlL8+nt913gSoMLh8nZVmXg45Rth5tZsWic377uvhXv\n/vxLnDFxOFaLK1BgNet8cawRh8OBpmkU/XIJa69ewLY7l2A2m5gyKtUvc8Z3Rr6jop5Z49IxaYLZ\n49LCUqV0DmK8uy8M9rphXIVrIHb0KroiMzmBZIuriCrFYmJKtpXi8nrXolVygmumUlbL1FE2/nx5\nPgWr3gfgT+tLmJuTzvaKevJzM3B0dLC1ojN80mF39lqbOyfNRGl9eOGWPSGEqBTxQ0KAwyQEhLJR\nc8elcu4jH9PY5sBq1vnHdQv59hMb+eFal7rq7juXceVfC/3EuOx2J+/tr2LptOGYTKYgLaoXrl7g\ndnAMFoTZayE/JzqrYIdCMTfujHysU9PS4W003GJ38tiKuWiaS/v6WEM7m0tqAPj0UAPz3Qbew+xx\naTx5WQGZyQmc98iHfq8tX72+12MJ18D3l4EUzFJEhife/8Jve864dIrL65gxJpXtFZ1hwJvPnsJ3\n12wGXI0/ikv8Qzyv7zzit252qL6F0x943/v67juX0ew0vIbdVyJYSklehoWS2u4zgwTw2Io5UTmT\nH4p1w7gK18QaoaZtWVYzBe7V94LcDEakWrxVfTJg6Stw4vq9heMYbrNQ09LBnkr/PptHm6PXiioD\nH/3sPtpEcoLGX78/j+23n0lRWR2GxM/AA4zL8BcdW14w2m/7W/mjmJubga4JZoxNY2e5f2jxgifW\ns2jVOlas2cSwFP+FSSEEDy6f3eNYZ41Li9rZvMquGQCitTKuq2lbd6vvWsD4k80aJ2eY2VnZxqxR\nyYzNSsXhMDje2MacsTaKK4KzXRSKvtLSYTDclsjfiw777R+bbuFIfTtzczMoC2jsvf1IE3vvPtsb\njtE0nQcvmsXKvxayvayOpz/yr9w+UOVq9N2Vl2tJ6F4kY/bYVP7vusVR9Vv3ZSiya+LayEdzx6ju\npm2edQXDkBxvavebss7Py/TG5F/94WKE0LxfGKdTdsobW0y8+5PTWflcEQeOt5Bi1oM69PSHRJOg\nrYuF1qnZyew71tLrNQBF9PDQRdPRENz08k7vPqtFJy3JxJmTMrnvvz7HXjyTdqdgfm4GFz7pL1lw\nclYKiYkmzp05CsOQLF+9gc0lnQ1otpY38OHPv8SuikaWTM3i0mcKu+0LMWWkDavF5JX8GJuqc8u5\n08jPySQhwRQTWjaDvW4Y10Y+mvPme2p00tUD6sWVwV6A5z0dqGrsVO1rd2B3wFs3f5nqZjsZSSa+\n/cQnbI9QGmWbQzJ9TCp7DjeSlKDR5PMAOXhcGfhYwqOL5MvfNpZTXNGA1a09M22UlSSTxqL73ws6\n/8KntgD41WaAq4NTdlqSdzYtpaQwoMOYBK59rpB/Xncqnx9v5fmr5lPb6ujSy9U0jW13LA2pMa8I\nTVwb+WjtGAU9T9u6ekD5egF2u9Ovq86kbCs2i8mVg+xW7fMc7xF6ihQzR1vZdaQRp5Q0uQtTPHY9\n0GCAK4e+XQXfo44kU2dtw/BknZ8unUBdm8EDb32GITu1Z3Yf6fm742vg54xL4x/XLUZKvM5Kfk46\n00fb2HHIP4y4p7KZKXe/BeBtsNOd4TaZNKaNDlazVIQmro18tGtOdzdt6+kBZbc7mXrXGzilKztl\nzz3nuMScRlopKq1jSrYV32zm/ccau+z41Bf2VTbh9LlcT+Z71TencvMruyN2f0XfCGy67vvgrWpx\ncsur+7zbnmpUDxa9s3F3t/fQBE9dVoCmaVQ1tnc6K2V15GUGyxb4EqgbH8tEy3pg3GfXxKrmdE+F\nXUVltd6sFKd0bVc32ykur8cAisvrvcUghiG567VdER2f58cugFljgn+QE7P8syyUge89b990asSv\nGWijfcNqORndhzLtTjhltEunaEQ3InjJZp1hbiEyj7Oia4KkBI3Pj7d0ew+bpXcNdqKVaNLRinsj\nH8t094CaPz4D3b1bF67trtKzqpvtbPUpSdeAFXNHhD0OAUwckexX5ej5tzXRxOWLc/yOH5OaQLsR\nWw/VaOSyZzZF7FrTRvVsOJ+5bK7fduC8b9ooG3vc+kSBSqi+tLQ7qHFr1HiclX/feBrNYUwD/vmD\nBTz09mccqm6IerGx7hgKtcmu6Fe4RgjxO+B8wA58Dlwppaxzv/YL4CpczsNNUso3+znWiBMt06m+\noOs6e+7x73QPhAxPZVnNTBtlY6d70dUAvjo7lxeKjoW89vTRqegCdh5uwJCuUMxnx1pY/9Mvc7Cm\nmYZWOze9tMMrFHXwmH+89lBDB9AR8tqK8DnaFLlitEXjM3uMq3/sLrTz4FmQTRBwyphUdh5uJNGs\n02J3MntsKtsCcuQ14TLqBQHhRU0TTB5pY05Oup+zEcjULAtLHv4YgEfeO0D+2FRe+eFpUZMR1xui\naT2wXwJlQohlwLtSSocQ4gEAKeUtQohpwFpgPjAaeAeYJKXs9lE+mJ2hojm9ciBwOJzM+tXbNLc7\nsSWa2HbHEi5+amOXolMaMHW0jV0+ypIf/PQMzn1kPS12wzWLEMJPHbC/BMaLAX51wRR+/Z/9tDlO\nvA4mgTHxgebe8ydz5+udMfkHL5rBL/+121uB7Uuoz+qhi2YwPDWRheOHBWnCAzidBhc+5RLRm5uT\nwR8umklJdTMnZ6VQ19LOuY9uCDpnyy+XRE1GXG8ZTCeyO4GyfnnyUsq3fDY3Ahe6//114EUpZTtw\nUAhxAJfBD/4Uh4hoTq8cCEwmne13LPNrpfbyDxZztK6VL6qbGD8smRvWbmN7RT2GdBmX3YcbmTYy\nhd1HXdWzX/n9h94YrlPCLUtP5oG3DvR5TIGGItBoCOA3/94bMlvnRKArAz9Yxn/NRwdDGngI/qwA\nfuzOqbdaTGy7Yykmk380WNc1/vGDxX6Gb2xmCgC/e2t/0PVmjrZGVUZcb4kWHa1IxuS/D3hKJMYA\n5T6vVbj3BSGEWCmEKBRCFFZVVUVwON0zFOXFQ42mCYZZXTF+h8Ng79EGdF2wIG8Yda1O9h6p91uI\nk+A18EBQ7vtv+2HgIdhQJAh/dUMJtJ6gBr47wjXwvfUdfb14cKVN9mVy29Tu6DJdt6t1pvvOn+y3\nfVIq/OuG02MujBqN9OjJCyHeAUaGeOmXUspX3cf8EnAAz/d2AFLK1cBqcIVrent+X4n29MpI4xue\nys9JZ9fhem9KZV8bQvT3wxqdauZwQ+eCVEfsrrP1yIWzs3llW+Wg3rM/f05dE0F9CcKlLxkyycnJ\nFP7iK/z6jX38+Mt55IxIj/vf5GDRo5GXUi7p7nUhxPeA84CzZGeA/xDgK0A91r0vqoiW6dRg4B+e\nqvPrEjVUttXXwMc7g23g+8sHPzmdBJMpSNp3XKqJ8oauF4QfXzGbc6eP8tOHh+DCvVBkpSXz4MVz\n+j94hR/9CtcIIc4Bfg5cIKX0TYB9DVguhLAIIcYDE4HN/bnXUBCrTcFD4Rueys/pf7XgT5aMj8Co\n4guzHj+e5xn/+yHpiQnMHedfA1Hd2n2w6OxpI0Ma+Kl3vcElazYx6fY3sEdQQ0nRM/2NyT8G2IC3\nhRDbhBBPAkgpdwEvA7uBN4Dre8qsiTaiqZghEvgWVz22Ir/nE3rgsfdK+j+oOMMew7IN/7x2nt+2\nIWFreR1/v+40nvt+Z9JGSzedmXQhqG7uCHKMQhXuKQaP/mbXTOjmtV8Dv+7P9YeSeMy+8YSnjjW0\n9fta7U6p9GjiiDHDUr3N4KGzwE7TBKdPGO7VROqOubnp3Lh2K4WltUwbncq/rluMyaR7C/c8Ehzz\nx2d4z/FoKvn2eFVElrjWrukP0VTMEGmG2yzMGWOl+FD/BMuUgQ+NCVcWQqwwe3QKVY1t/PabM3l5\naykdTsmVi/O8RlfTNN66+QwWddF675RRqTx9xVw0TWPhqnUY0tW5bPav3mbbHcswmUIX7jkcRqc0\ntluYLDDtUtF/lJHvgmjOvmlrc/j1xOwtQgj+cf0ZlNU2sWL1Jg7Xd99OTRE+oWR7o51th5v52qMf\n++17ZkMF+WNtvPLD09E0wcjURGaOCVaQBPjlV6cwMj0ZKSXTRqfyqbuyuqnd6RUb0zRBRkCnpwNV\nTZ3S2HEkTBZtqMdmN0SjuFlbm4Mpd7/JdS9sZcLtb9LW1jefUdMEY9NSyEiOnxlKNBALBj7c9eGt\nFY1ezRUhBP/84WnMGmPzO0YDTh6RgpQSIQT/um4xVovLU/fIXXs89nMe/ohZ976Nw129PCnbii3R\n5HesIvIoTz7GeG9/VdD2uTNHhXWubxqbEBoXrd7AriOqRWAs0tfaBgi/p27+WJtfmFJKWPXtWQxL\nMZOemEBhWS0Pr/uMUx943ysNAoKXrl2EkNLtwWvsr2wI6bFrmkbx7UtVTH6AUUY+BNEsXLZ02vBu\nt33xfR8dHYaf/vyHt3yFbd2IRSmih6QEjdaArJZIroackp3EredNZ4zVwn93HSEtOYEzp2QzKiPF\n+/0PFT+fNDKVrWV13uSEyoY2lj30od8xmtbpsXv2+3rsJpOmQjQDjDLyAUS7cJnJZPJrjNxVTD7w\nfdx05gS/NLaDVU2kuH94iugm0MBHml2VrVz29Bbmub/vta0dQQ5OYPz87d2VLDtlhF9yQl2LPa48\n9mh29nqDMvIBxELqpKcxcncEvo8JI1L80tgmDLf6iU/pmkCXEnsMJcz4tq6LR8LtxBQptpTW8p2n\nPmHnoQY/B8dud1Ld3E5Kgkaz+4Fz3QtbsSWaKLptCfXtDtITTXx2rNHbEzbWPfZod/Z6gzLyAcRL\n6mTg+xiRmuSXxqZpGgW5GWw5WMO00TbWXF7AvqP1XPHnrUM99LDpi4E3axDBLogDxuPLZ7PslBFM\nuuOtQZWd2O7uLOZxcNIsJm+YTxPwyCWzuGntdsDlrX9R3cyE4VZvKMdq1vn3TacxdaQtZjz2UMSC\nsxcusfspDBA9td2LFUK9D7NZZ9GELHRdRwjBX6+cT5JF59PDjSx96CMWjs/qk+rgQJEYWuIkbGaN\nSQ1SYhyWYuL5K2eFdb55CH8df95QgpQaln7+DXpDUoIWpMzqW61qSBiWZA7KiPEN5TTZnehCdGng\nY0UqJJ5UapUnH4J4ES7r6X18Ud3sbcnW2Obg3X1V7L5rGcUV9RTkprP7aCNff/yTwRpuEP1ORxSC\n607P5Y8flXp3HWl08N1nt/d4qk7XHv9NXzmJR977ol9De/2GU6lrtXPZ01tCvl5UVsfm0upBTcls\n7TB47LsuyQvDMNh7pIGC3DS/MN/Ck4cFxde7W1j1JZZCINFcJ9NblJE/gZmUbfUrV7/uhWJsFhPF\ndywF4FBN61AOr99sr6hne0V9yNdMArqL9gTaVt/4f38NPEB5dQtzuhGKMyT87o29YV0r0aRFpHPW\ntJFWslLMLF+9kS2lLn0Zm8XEzjuWsf1wPfPHZ9DRIXlv/zGWThvuVxEbzsJqrIVA4sbZG+oBxBKx\nMtUMF0+5ui+N7Q42HDzOnF+9zQ/XFg/RyIKxaEQ0lOQx8GbhCut0hyag+Pal3H7ulIjd/4driznt\nt+8HqTxOG2n1hpg+PdzIrLGdr6eYNUZYE/yO14HtdyxhysjQ3rMgvOYhyQkar994GjUtHX4CYo3t\nDsrqW1k0IYuODtllIZ5nYbW7OHw8hUBiCWXkwyTeVCk9jExNZF5uundbF3DZ01t6FKMaLJJMgocu\nnM6EbGtQZ6pIYJfwm2/NCKrk9MWQ8LdNZXx3/ug+3yeUoXVK+J+zpzBnXDqagNlj03jme/MoyE1H\nF5Cfk45Z19GEq7l68e1LWfXNaX7XeGVlAQkJJv585fyQ7yHRJLpduNWF4Pmr5/PpPWej67rXEHvw\nbQASqhCvN0TTele8OWzdocI1YRJrU81wEULw0rWLqWxs42BVE5eu8Zf9TzbrXfb57IoMi6C2PfjH\n85Mlefz+nZJeXavVIfnxK5/26pxQzBht49PDjUEGL8kkmJCVwufHuw9N3fffvdz337385ep8Ll/T\n+wykBF2ElCJOTzSxvbwOA9hWUc/iB94jxWwCIehwGmyrcOnAfHq4gYtWb2CnT2N1iwYzcrL8On6t\n//mX+WD/MX75r91AV+jTCwAAFoVJREFU1xlIyWaddodBQW4Gi0/O8hpcIQQvrVxEZUMbdS12Jvtk\nyfSmEK8roiEEEktrA5FAefJhEutTzUDPxXdb0wSj0pJYdHIW88ZnYtIE83Iz+O9Np7HzrqXMz8tA\nF6LLkEAgJ2en8uFPv8SUkTZ0TTBxuJX99y5jybSxA/kWu2VngIH/2dKTmZydQrtTcsHjH9MU5sxl\n02fVfbp/KAM/ZUQy5z32iV/PVkO6QiROQ3oNvIdtFQ04DYmuCdZevYCdd5/NlpJatpTUeJ0Ps0nn\nzMnZPY5n6kgrn9xyptejttudbDhwHKfT6fo+pCcxdXSaX/jFU4j3xIp8Dtx3dp/E8aKBUA5bPBOb\nn9IQEMur7YGey/NXLeC7T28K8mS6eo8vrlxEdbOdzGQTs+99p8dQzraKBlo6DA4ca8JpSL6obubz\n4y1+C72JJsH0MWkUlg6NtMIj6z6n3W1d9x1r7v5gH/74QWnPB/nw82UTumx4vvdYS8j93aEJKMjN\nYPooGzPvfZM2R6en5pRww9piHlk+2++cOePS2XGonhmjbew41IAhYXtFg/cz93Ru8mTQ7LnnHEwm\njaP1rZRUN7NgfKZXHjicQrxoJ15qYcJFGfleEA1Tzd5iGJLdR+rZUlKDIaGwtJbPjjVSWFKDU0Jh\nSY1f6CnUe/TdV3yHK4si1aJz+u8+8PaK1QSkWEy0tDuYm5vBpGwr+TkZbC6pwWlIzn1kPfPzMtny\ni7O48E8b2H24AU0IXr/hVC547OOQcWOzgBFpFirq+i6F3FXxU3svklGSEgRtHZKJw5PZX9U7w9yV\nge8rE7OSWDY1gxm/ese7z8D19zckbC2tRROCebnpFJXWMTc3nRdXLqKmpYNhKQlc8qdNQcYtsHPT\nlrIaHnp7P1vcD2BNwN57zsFs1uOiyUcsO2x9QRn5OMYwJBc/tcGbDgfgNCR3/GsniQk6zXYnyWYT\nHU4nxxrawpJV9mRRSCkpyMtwx4IzeGzFHIalmKlp6dQ9eXTFHBb9Zp03HFFUVktJbQt7jzRiSFdV\n5Z2v7vIa+Hm56fxoyUQudeeO2yX9MvAQbOCnj+nUOw+X1g7Jv286je88MXQ1Ax72VbXyq/987rcv\nAZjj9kxnjk0jMzkBITSEJhCahhCdD+m11yykqrEdp3TlwU8eaQvq3HRyVgpFPjMsw92yb15eZtw0\n+YhFh62vKCMfx1Q320P20yws68wdb2x3sPj+9wCYPTaVX39zBlN7SIWDrr0h3x/OCJuFgrwMNpe4\nxuDx8Of6GCRPHruuCe79+nQmZVtJsejeIq1IMj/PFZraX9nIuY+sD/u82ePS0Oi+v+lQouvwl+/N\n55JnNrG9op7vrN7IjvI6nG7PPjBJ4IYXirxeuqcuIlDyYm5uup8nP398Bp8dU00+YhFl5OOYzOQE\nLDq0hpkNua2iga89+jFWs862O5f16KV15Q0dr2/hN2/s477zJ/PiykVUNbUjwDtTWHvNQqqa2kFK\nbnxxG0WltSSbdc57dD0FeZkU3noWZz34AYcbItexSgD3XHAKmiaYMiq1yy5HgZwyysYr17rWJJIT\ntKg09G1OKK6oY2dFPU5DsqOinlnj0tlRUR8Uc3YtOnZ66Y3tncZ60YQs7/6Xrl0cFJMPt7JVEV0o\nIx/H1LR0hG3gfWmyO9lX2cDJw6y9bjN4vL6FglWumcH/FR9m861fRtNNQbHPm9YWe9P+Xr/hVM5/\ndL1rjaC0lhXPbIqogQewJpr46iPrmTzSyuvXn8o/f3ga3/rjx2zvJnQza4yN33xrJpes2URxWR2T\nsq3sjsImK1aLq1m272LiC1cv8AudeXAtOnZ66V0Za00TjLAl0tDm8J4fq5LBJzoimooBCgoKZGFh\n4VAPI26QUvLVhz9gz9Hws0c8BMrc7r37bBITezb0N79UzP8VH/ZuD0tOoL7N4ZfFU9XYzqJV63AY\nEpMm+OTWM7nRbfRnjE1je1kdkfSXn1iRz3UvdOa2awJ233U29XYHSMnnVU2sCKgPWP+zMzj3kU+6\nzCRKThA8cVk+VzxT5N03PMVEVfPgFZHNHG3jlnOmsPDkYei6Hrb+uWFIKhvbqGv2z4P3RTXZji2E\nEEVSyoJQr6lPLY4RQvD6Daf3Sc0xMCQebnXjfedP9tuubukIykcOrDkYbrN4KyH/8YNFFORlhLp0\nr/CYOJvFxJKpw0lK6PyqGxK+/sR6Fq9ax41ri5mfl+lVVkw0aey/dylNduln4ANN5ls/OoOH3vbP\nnBlMAy8ATde44s+FrFizGcOQQT2JHQ7X4qph+D8yPXURgXnwvoRqsq2ITVS4Js4xmXQ+vedcDlQ1\nkZueyFcfX8/BHqo7Q3HWlKyeDwISE5OYkZ3Izso2po+wkJycwtYy/5S9UIu2QnQu2nri+EhJU5ud\nv20o4dmNFb0ab/64VIorGpgyykZNi8OvGEkA+ypds5vNJbVUt3RQdNsSispqKchNp67NycQRKd74\nMwS321v5t0J292GGFMiqb0zjF+7q1FAk6p1qnHNz0nBK2HmogVnuRWtnFxXYXXni4aZAqvh7/BAR\nIy+E+Anwv8BwKeVx4XIlHga+CrQA35NSxk43ijjDtyvPO//zZb75+Hp2HO5dbLm+3WB4GDUj1c12\n9lS54ul7j9v5+NbT0ITwCyHY7U62lNVwclZKyGtomiA7NRHDkNz04jYKS2v9DG5XWDQYnZ5IWV07\nReWuWHtxWR3CXUBUWFrLxBFWbjt3Epc/2xlmQUoue3Yzmw/WkJigYXdKV3z7tiWs23vML9TjoT8G\n3hMKsyWa+E7BOH793/00tTvQhKtZ9uSRVp6+fC4JJhPDUswcb7J7H4JSuv7GXeW8ewjlifs29+gp\nBKPi7/FDv428EGIcsAwo89l9LjDR/d8C4An3/xVDjK5r/OuG09lWUsO3Vm/s9tgJWUkcrG7tVVVg\nUEeqgNx7u93JlLve8IqNzctN56VrF4fUDvGUnzsNSYvdyQtXzw+KnXuYnJ3Cc1fO57Tfvo/TfXFd\n4BcOOtbQxpIHP+DyZ4tc6owCCvIykRK2lNQg6eynuvlgDbWtds6enu1taZdoErT1oRtVijsrR+Ka\nRey4cxkHa1q9xnObu8BswvAUalsdQTH17LRE7799ZzzdFfSE8sT3V/YuBTLWWvYpQhOJx/ODwM/x\nn9F+HfiLdLERSBdCxHYtdByhaYI54zOZ5xP7tlp0nrncvxzebNJdVkUIwl2f70lp8P+3d/8xctRl\nHMffn7v2CvToHaW0Pbg7CqGl9geEcjRFLJEfYkXgDDFaNYpIqJRSa4JRSv0ZNeKPgBjEpAEMJJha\nEKUigoDEIFJKf1jgALVRCyXFitoiFq/29vGPmb1ur7e7Mzu7tzN7z+uf3s7O7n6/O91nvvPMd57Z\n9PK/DqomuWn77qK1Qwpz9z3HH8WCE49m/jD5+lkdrfxyxdlMaTt8cP350yYe1IamJrH7rf/xZniy\nwYBxY5q48QOnsHzNlmErXF555zPkcjCrYwLNEnM72yP9YATMmnrgKGXm1Fb6vngBN35wLl9939v4\n597+g0545oPpmDHNkS5Iy8vn4AcG7JDce34k/tCKhWz94rsOurkHFJ9V4xpPotk1knqBc81shaS/\nAj1huuYB4AYz+2243mPA58zskKkzkpYASwC6u7tP3749Xm0QV7lczg6qNggw98u/Cq6EHRvciCJn\nMKZJPLXyvKpcITgwMMD0zx88kl971duLBrahM0ZyOeP1N/u5+u5NbN6+m1M627jv6rcPBsxSM0xy\nuRxzvvRwrLnuD37qHVx8y5ODhcHmdLSytcz8+tnHHknfkHTYnI5Wnt954OTl6V1t3LP0rMTVD+PO\ngmmEsgTuUKVm15QN8pIeBaYO89Qq4HrgAjPbU2mQL+RTKOsvHwSmTx7Ph2/bMJh2qWb978Kc/JS2\nwyt636jTBYfa/PI/uPTW0mmqQs3AuJYm9u7LMX5cM2/1D5Sd3vnzqxdw3U+fp29n6Rkpz6w6v+yO\ns1w/X9r5BotufmLw8UMrFnqKZRQqFeTL5uTN7PwibzoXOAHYGv7n6wQ2S5oPvAp0FazeGS5zKVeY\nh61VEaeWlmbOOil+LfJCldQeyeWML9/fF+s1A8DesADO3v4BTusOriSde9wEtu7YM2ya5+Jb1zOv\nawLjW4L6QK3jmjl58vjBk8EAPd1tZc9zRKl77rNgXDkVn3g1s+eAyfnHQ0by64BrJK0hOOG6x8x2\nJm2sG1n1KuIUZ5QeJ/3wj//s47kIpQyKOa27nXs+eaCi4+LV6wfr8syc0spLfzswct/yyhv87rpz\n2fPW/8LAK3b88z88ue113nnyJKa2jy/btyg3qvFZMK6cWs2Tf5Bg+uQ2gimUl9foc1yDiXPXnrj5\n6EmtLfQUXNIfR2fbOO696kyampoOmc8vYOIRYzj5Cw8Pluw9/fh2prYdRkf74YPv0T2ple5J0Ufa\nUeue+ywYV0rVgryZTSv424Bl1XpvN3rEuc3icHPBZ3ZMKHokkL/V4WtvvMWVd20cPDk6u+NI+srU\npPn1tWcfMkrOz+fPe/ErixKfayg02uqeu9rwK15dqsS5a89w+ehyRwJNTeLY9iP4+TULC26HaLzr\nu0/w7//u54iWZi5f0MUZ09o5vftoHuzbxaXzOhg7dmzZtlfjXMNQo6nuuasNL1DmUidJTn5o8bNS\nUz8LdwindbXxld45zCxSsMu5NPMCZS5ThhbaKiWfj84H5jg3XC9MDW15ZQ/HHHmYB/gRMvTG8q52\nPF3jGkqcPPZou6FzWsQ5ue6S8yDvGk7UPLaf2KyPOCfXXXJ+bOpGtTipIVdelDRMnJSaS85H8s65\nqoiahvEjqJHlI3nnXGSlRurDpWGK8SOokeNB3rlRptKZLfmR+pnfeIzFq9eTG1K4x9Mw6eTpGudG\nkSQzW8qdMPU0TDr5SN65USROSmWoKCN1T8Okj4/k3YirtBa8Sy7JtQE+Us8mD/JuRPmFMPWVNFB7\nLZ3s8XSNG1FJ0gWutKgnVD2lMrp4kHcjymdg1Ea5mS9RXu+1ZBqTp2vciPK8bm0kKRXgKbTG5iN5\nN+LSnC7I6og2yRGSp9Aam4/k3YhJ+6yaNI1o435XSY6QvBpnY/Mg7xKJGozSFECLqUV1xEp2bJV+\nV0NnvkT9bE+hNTZP17iKxTnZV++UQLWrI0Z5v0pPhlbju4r72WlOoblkfCTvKhZn5BsnJVBpWqfY\n66pdHTHq+1V6ZFDuu4ry/XjNdpfnQd5VLE7grnYAjfO6OAEvysU+Ud+v0mBd6ruK+v14nt3leZB3\nFatFLrfSEWip11U74EV9vyTButjOJur343l2l+dB3iUS9TL3ao1Ai41+S72u2gEvzvslDdZDxdlh\neQkCBx7k3Qipxgi01I6iXOCNszOKklJKusOo9OjCR+gursRBXtJyYBkwAPzCzD4bLl8JXBEu/5SZ\nPZz0s1x2VWMEWm5HkXTkGuVoI+oRSbkdwdBgbQavv9kfKXD7CN3FkSjISzoH6AVONbN+SZPD5bOA\nxcBs4FjgUUkzzGwgaYNdNlVjBFqN3Hqp4BvlaCPKOlF3BPlgnYVrCFx2JR3JLwVuMLN+ADPbFS7v\nBdaEy/8iaRswH3gq4ee5DEs6Ai23oyg3ei4XTKPsRKKsEzff7tMdXS0lDfIzgIWSvg78F/iMmT0D\nHAesL1hvR7jsEJKWAEsAuru7EzbHNbpiO4ooo+Fq3L4uyjpxjzh8uqOrpbJBXtKjwNRhnloVvn4i\nsAA4A1gr6cQ4DTCz1cBqgJ6enmxVhXKpEWU0HCWYltqJ5AN7uSOSuKkpP5nqaqlskDez84s9J2kp\ncJ8F13ZvkJQDJgGvAl0Fq3aGy5yriSgBvNJgWknOPG5qyk+mulpJmq75GXAO8LikGUAL8DqwDviR\npBsJTrxOBzYk/CzniooawCsJpp4zd1mWtEDZHcCJkp4H1gCXWaAPWAu8ADwELPOZNa7WalVky+9m\n5bJMabo5Qk9Pj23cuLHezXDuEGmvhe9GN0mbzKxnuOf8ilfnIvCcucsqryfvnHMNzIO8c841MA/y\nzjnXwDzIO+dcA/Mg75xzDcyDvHPONbBUzZOX9Hdge4yXTCK4wrYReF/SyfuSXo3Un6R9Od7Mjhnu\niVQF+bgkbSx2AUDWeF/SyfuSXo3Un1r2xdM1zjnXwDzIO+dcA8t6kF9d7wZUkfclnbwv6dVI/alZ\nXzKdk3fOOVda1kfyzjnnSvAg75xzDSyTQV7SckkvSeqT9K2C5SslbZP0B0nvrmcb45J0rSSTNCl8\nLEnfC/vzrKR59W5jOZK+HW6XZyX9VFJ7wXOZ2zaSFoXt3Sbpunq3Jw5JXZIel/RC+DtZES6fKOkR\nSX8K/z2q3m2NSlKzpC2SHggfnyDp6XD7/FhSJu7mIqld0r3hb+VFSWfWcrtkLshLOgfoBU41s9nA\nd8Lls4DFwGxgEXCrpOa6NTQGSV3ABcDLBYvfQ3DbxOnAEuAHdWhaXI8Ac8zsFOCPwErI5rYJ2/d9\ngu0wC/hQ2I+s2A9ca2azgAXAsrD91wGPmdl04LHwcVasAF4sePxN4CYzOwn4F3BFXVoV383AQ2Y2\nEziVoE812y6ZC/LAUuAGM+sHMLNd4fJeYI2Z9ZvZX4BtwPw6tTGum4DPAoVnwXuBu8LbKa4H2iV1\n1KV1EZnZr8xsf/hwPcEN3CGb22Y+sM3M/mxm+whub9lb5zZFZmY7zWxz+Pe/CQLJcQR9uDNc7U7g\nffVpYTySOoH3AreFjwWcC9wbrpKJvkhqA84Gbgcws31mtpsabpcsBvkZwMLwMO03ks4Ilx8HvFKw\n3o5wWapJ6gVeNbOtQ57KZH8KfAL4Zfh3FvuSxTYPS9I04DTgaWCKme0Mn3oNmFKnZsX1XYKBUC58\nfDSwu2BQkZXtcwLwd+CHYerpNknjqeF2SeXt/yQ9Ckwd5qlVBG2eSHAIegawVtKJI9i82Mr053qC\nVE0mlOqLmd0frrOKIF1w90i2zR1KUivwE+DTZvZG4f1pzcwkpX4OtaSLgF1mtknSO+vdnoTGAPOA\n5Wb2tKSbGZKaqfZ2SWWQN7Pziz0naSlwnwUT/DdIyhEU93kV6CpYtTNcVnfF+iNpLsGefWv44+sE\nNkuaT0r7U2rbAEj6OHARcJ4duAgjlX0pI4ttPoiksQQB/m4zuy9c/DdJHWa2M0z/7Sr+DqlxFnCJ\npAuBw4AJBHntdkljwtF8VrbPDmCHmT0dPr6XIMjXbLtkMV3zM+AcAEkzgBaC6m3rgMWSxkk6geCE\n5Ya6tTICM3vOzCab2TQzm0bwH2Cemb1G0J+PhbNsFgB7Cg7nUknSIoJD6kvMbG/BU5nbNsAzwPRw\nBkcLwYnjdXVuU2Rhzvp24EUzu7HgqXXAZeHflwH3j3Tb4jKzlWbWGf5GFgO/NrOPAI8D7w9Xy0pf\nXgNekXRyuOg84AVquF1SOZIv4w7gDknPA/uAy8IRY5+ktQRf2H5gmZkN1LGdST0IXEhwknIvcHl9\nmxPJLcA44JHwyGS9mV1lZpnbNma2X9I1wMNAM3CHmfXVuVlxnAV8FHhO0u/DZdcDNxCkOK8gKOv9\ngTq1rxo+B6yR9DVgC+HJzAxYDtwdDh7+TPDbbqJG28XLGjjnXAPLYrrGOedcRB7knXOugXmQd865\nBuZB3jnnGpgHeeeca2Ae5J1zroF5kHfOuQb2f20e0pAu67r0AAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "g7HUx_4_1P7H"
      },
      "source": [
        "There is a big cloud of data but we can see small groups that form within it. This does make sense because we are sampling from a dataset that is billions large. The small clusters that we can see reprsent the groups of words that are similar."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MT29e2yN0QmA"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}