LSTM.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "LSTM line speaker.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "O-Rb4DXeVWoH"
      },
      "outputs": [],
      "source": [
        "# lets try a long short term memory neural network. \n",
        "# why this and not a simple RNN? because our data is sequential and the data has a temporal relationship\n",
        "# meaning, each sequence (line) has a relationship with the sequence before and after\n",
        "# they arent just loose datapoints, they are very much related\n",
        "# also, simple RNN cant handle long sequences, and we know from the last models that the longest sentence is more than 200 words\n",
        "# LSTM can handle this, i believe"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import tensorflow as tf\n",
        "import tensorflow.keras as keras\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import seaborn as sns\n",
        "import pandas as pd\n",
        "\n",
        "# to get reproducible results:\n",
        "np.random.seed(1234)\n",
        "tf.random.set_seed(1234)"
      ],
      "metadata": {
        "id": "q1pG8GI-WLz-"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from google.colab import files\n",
        "uploaded = files.upload()"
      ],
      "metadata": {
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "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",
              "ok": true,
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "status": 200,
              "status_text": ""
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 91
        },
        "id": "8b667-9BWL2O",
        "outputId": "065d5cda-3c84-450e-c17e-f4d2c759c84d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-1be77d7b-8401-4ece-ad4f-f0cccea7527e\" name=\"files[]\" multiple disabled\n",
              "        style=\"border:none\" />\n",
              "     <output id=\"result-1be77d7b-8401-4ece-ad4f-f0cccea7527e\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Saving The-Office-Lines-V4.csv to The-Office-Lines-V4.csv\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import io\n",
        "df = pd.read_csv(io.BytesIO(uploaded['The-Office-Lines-V4.csv']))"
      ],
      "metadata": {
        "id": "4FN3inXwWL4G"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "df = pd.read_csv('The-Office-Lines-V4.csv')"
      ],
      "metadata": {
        "id": "C0urctHAWL6U"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "df= df.drop('Unnamed: 6', 1)\n",
        "\n",
        "def gender(name):\n",
        "    if name == 'Michael' or name == 'Jim' or name == 'Dwight' or name == 'Ryan' or name == 'Stanley' or name == 'Kevin' or name == 'Oscar' or name == 'Toby' or name == 'Creed' or name == 'Andy' or name == 'Roy' or name == 'Darryl' or name == 'Gabe' or name == 'Robert California' or name == 'Robert' or name == 'Clark' or name == 'Pete':\n",
        "        return 0\n",
        "    elif name == 'Pam' or name == 'Phyllis' or name == 'Angela' or name == 'Kelly' or name == 'Meredith' or name == 'Jan' or name == 'Erin' or name == 'Nellie':\n",
        "        return 1\n",
        "    \n",
        "df[\"gender\"] = df[\"speaker\"].map(gender)\n",
        "\n",
        "df = df.dropna()\n",
        "\n",
        "df['gender'] = df['gender'].astype(int)\n",
        "\n",
        "df['number_of_words'] = df['line'].str.split().str.len()\n",
        "\n",
        "df['speaker_id'] = pd.factorize(df.speaker)[0]"
      ],
      "metadata": {
        "id": "95l0zVBWWL8f"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "df"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "EVgsNa0kWL-W",
        "outputId": "caed84a7-decd-4032-dd23-876538bc3e80"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "\n",
              "  <div id=\"df-41ec1295-cdfb-461c-bb64-2aaaa2f17787\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>season</th>\n",
              "      <th>episode</th>\n",
              "      <th>title</th>\n",
              "      <th>scene</th>\n",
              "      <th>speaker</th>\n",
              "      <th>line</th>\n",
              "      <th>gender</th>\n",
              "      <th>number_of_words</th>\n",
              "      <th>speaker_id</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>Pilot</td>\n",
              "      <td>1</td>\n",
              "      <td>Michael</td>\n",
              "      <td>All right Jim. Your quarterlies look very good...</td>\n",
              "      <td>0</td>\n",
              "      <td>14</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>Pilot</td>\n",
              "      <td>1</td>\n",
              "      <td>Jim</td>\n",
              "      <td>Oh, I told you. I couldn't close it. So...</td>\n",
              "      <td>0</td>\n",
              "      <td>9</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>Pilot</td>\n",
              "      <td>1</td>\n",
              "      <td>Michael</td>\n",
              "      <td>So you've come to the master for guidance? Is ...</td>\n",
              "      <td>0</td>\n",
              "      <td>14</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>Pilot</td>\n",
              "      <td>1</td>\n",
              "      <td>Jim</td>\n",
              "      <td>Actually, you called me in here, but yeah.</td>\n",
              "      <td>0</td>\n",
              "      <td>8</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>Pilot</td>\n",
              "      <td>1</td>\n",
              "      <td>Michael</td>\n",
              "      <td>All right. Well, let me show you how it's done.</td>\n",
              "      <td>0</td>\n",
              "      <td>10</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>54621</th>\n",
              "      <td>9</td>\n",
              "      <td>24</td>\n",
              "      <td>Finale</td>\n",
              "      <td>8153</td>\n",
              "      <td>Creed</td>\n",
              "      <td>It all seems so very arbitrary. I applied for ...</td>\n",
              "      <td>0</td>\n",
              "      <td>55</td>\n",
              "      <td>16</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>54622</th>\n",
              "      <td>9</td>\n",
              "      <td>24</td>\n",
              "      <td>Finale</td>\n",
              "      <td>8154</td>\n",
              "      <td>Meredith</td>\n",
              "      <td>I just feel lucky that I got a chance to share...</td>\n",
              "      <td>1</td>\n",
              "      <td>41</td>\n",
              "      <td>14</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>54623</th>\n",
              "      <td>9</td>\n",
              "      <td>24</td>\n",
              "      <td>Finale</td>\n",
              "      <td>8155</td>\n",
              "      <td>Phyllis</td>\n",
              "      <td>I'm happy that this was all filmed so I can re...</td>\n",
              "      <td>1</td>\n",
              "      <td>31</td>\n",
              "      <td>5</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>54624</th>\n",
              "      <td>9</td>\n",
              "      <td>24</td>\n",
              "      <td>Finale</td>\n",
              "      <td>8156</td>\n",
              "      <td>Jim</td>\n",
              "      <td>I sold paper at this company for 12 years. My ...</td>\n",
              "      <td>0</td>\n",
              "      <td>46</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>54625</th>\n",
              "      <td>9</td>\n",
              "      <td>24</td>\n",
              "      <td>Finale</td>\n",
              "      <td>8157</td>\n",
              "      <td>Pam</td>\n",
              "      <td>I thought it was weird when you picked us to m...</td>\n",
              "      <td>1</td>\n",
              "      <td>46</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>47664 rows × 9 columns</p>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-41ec1295-cdfb-461c-bb64-2aaaa2f17787')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-41ec1295-cdfb-461c-bb64-2aaaa2f17787 button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-41ec1295-cdfb-461c-bb64-2aaaa2f17787');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ],
            "text/plain": [
              "       season  episode   title  ...  gender number_of_words speaker_id\n",
              "0           1        1   Pilot  ...       0              14          0\n",
              "1           1        1   Pilot  ...       0               9          1\n",
              "2           1        1   Pilot  ...       0              14          0\n",
              "3           1        1   Pilot  ...       0               8          1\n",
              "4           1        1   Pilot  ...       0              10          0\n",
              "...       ...      ...     ...  ...     ...             ...        ...\n",
              "54621       9       24  Finale  ...       0              55         16\n",
              "54622       9       24  Finale  ...       1              41         14\n",
              "54623       9       24  Finale  ...       1              31          5\n",
              "54624       9       24  Finale  ...       0              46          1\n",
              "54625       9       24  Finale  ...       1              46          2\n",
              "\n",
              "[47664 rows x 9 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.model_selection import train_test_split"
      ],
      "metadata": {
        "id": "BxW56qteWMA7"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "X = df['line']\n",
        "y = df['speaker_id']\n",
        "num_classes = len(pd.unique(df['speaker_id']))\n",
        "\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, shuffle=True, random_state=42, test_size=.33)"
      ],
      "metadata": {
        "id": "_D--u1N7WMDC"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "X_test, X_val, y_test, y_val = train_test_split(X_train, y_train, test_size=0.5, random_state=42)"
      ],
      "metadata": {
        "id": "84FyuBaBWMGC"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(len(X_train))\n",
        "print(len(X_test))\n",
        "\n",
        "print(len(X_val))\n",
        "print(len(y_val))\n",
        "\n",
        "print(len(y_train))\n",
        "print(len(y_test))\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "N40C7c_RWnCt",
        "outputId": "bf7d4a64-a9ff-4740-b413-bf64fd294219"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "31934\n",
            "15967\n",
            "15967\n",
            "15967\n",
            "31934\n",
            "15967\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# padding: not all lines are the same length, but they have to be the same length for NN; padding will fix this\n",
        "# do this for train, test and val sets\n",
        "from tensorflow.keras.preprocessing.text import Tokenizer\n",
        "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
        "\n",
        "lines_Xtrain = list(X_train)\n",
        "lines_Xval = list(X_val)\n",
        "lines_Xtest = list(X_test)\n",
        "\n",
        "# there will be a max length of line (maxlen), which should be the length of the longest sentence. this number should be the same in all sets\n",
        "longest_line_Xtrain = (max(lines_Xtrain, key=len))\n",
        "maxlen_Xtrain = len(longest_line_Xtrain.split())\n",
        "\n",
        "# longest_line_Xval = (max(lines_Xval, key=len))\n",
        "maxlen_Xval = maxlen_Xtrain\n",
        "# len(longest_line_Xval.split())\n",
        "\n",
        "# longest_line_Xtest = (max(lines_Xtest, key=len))\n",
        "maxlen_Xtest = maxlen_Xtrain\n",
        "# len(longest_line_Xtest.split())\n",
        "\n",
        "# only 1000 most frequent words & oov_token so any unknown word in test set wont just be thrown away, but will be replaced by token we choose\n",
        "tokenizer = Tokenizer(num_words = 10000, oov_token=\"<OOV>\")\n",
        "tokenizer.fit_on_texts(lines_Xtrain)\n",
        "word_index = tokenizer.word_index\n",
        "\n",
        "sequences_Xtrain = tokenizer.texts_to_sequences(lines_Xtrain)\n",
        "sequences_Xval = tokenizer.texts_to_sequences(lines_Xval)\n",
        "sequences_Xtest = tokenizer.texts_to_sequences(lines_Xtest)\n",
        "\n",
        "\n",
        "# default padding is pre, so zeros will occur before ints\n",
        "padded_Xtrain = pad_sequences(sequences_Xtrain, maxlen = maxlen_Xtrain)\n",
        "padded_Xval = pad_sequences(sequences_Xval, maxlen = maxlen_Xval)\n",
        "padded_Xtest = pad_sequences(sequences_Xtest, maxlen = maxlen_Xtest)\n",
        "\n",
        "print(\"\\n Padded sequences train: \\n\", padded_Xtrain)\n",
        "print(\"\\n Padded sequences val: \\n\", padded_Xval)\n",
        "print(\"\\n Padded sequences test: \\n\", padded_Xtest)\n",
        "\n",
        "# now each row in the matrix is a line with the same length, and oov_token specifiec out-of-vocabulary tokens that arent recognized\n",
        "# also, only 10 000 most frequent words are included\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cJPhcJkaWveM",
        "outputId": "5157f109-43df-42de-f8d7-1e5e063fc98d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            " Padded sequences train: \n",
            " [[   0    0    0 ...    3  986   17]\n",
            " [   0    0    0 ...    0    0 8198]\n",
            " [   0    0    0 ...   78  668  213]\n",
            " ...\n",
            " [   0    0    0 ...   38   12  594]\n",
            " [   0    0    0 ...   19  145 4544]\n",
            " [   0    0    0 ...   35   34 1509]]\n",
            "\n",
            " Padded sequences val: \n",
            " [[   0    0    0 ...   25   14   68]\n",
            " [   0    0    0 ...   27   23  591]\n",
            " [   0    0    0 ...   46    7 1343]\n",
            " ...\n",
            " [   0    0    0 ...  651  704  690]\n",
            " [   0    0    0 ...    2   76  232]\n",
            " [   0    0    0 ...    9   56   15]]\n",
            "\n",
            " Padded sequences test: \n",
            " [[   0    0    0 ...  584    3   68]\n",
            " [   0    0    0 ...    0    0 3819]\n",
            " [   0    0    0 ...  289  180  455]\n",
            " ...\n",
            " [   0    0    0 ...    0    0 4679]\n",
            " [   0    0    0 ...    0    0   45]\n",
            " [   0    0    0 ...    0  391 5887]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# data should be float32, not int. goes for output too\n",
        "# also everything should be a numpy array\n",
        "y_train = np.asarray(y_train).astype('float32')\n",
        "y_val = np.asarray(y_val).astype('float32')\n",
        "y_test = np.asarray(y_test).astype('float32') \n",
        "print(y_train.shape)\n",
        "print(y_val.shape)\n",
        "print(y_test.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "top-ZSKBW0ml",
        "outputId": "e3432ff5-9bda-4b1a-fbf6-a0ce15481dda"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(31934,)\n",
            "(15967,)\n",
            "(15967,)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from tensorflow.keras.utils import to_categorical \n",
        "Y_train = to_categorical(y_train, num_classes=num_classes)\n",
        "Y_val = to_categorical(y_val, num_classes=num_classes)\n",
        "Y_test = to_categorical(y_test, num_classes=num_classes)\n",
        "\n",
        "print(y_train.shape)\n",
        "print(Y_train.shape)\n",
        "\n",
        "print(y_train[:10])\n",
        "print(Y_train[:10, :])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VXkXBSXdZ3cY",
        "outputId": "df2bf66b-f976-4fcc-946a-a589e708e065"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(31934,)\n",
            "(31934, 25)\n",
            "[ 3.  1.  3.  3.  0.  9.  0. 17.  0.  1.]\n",
            "[[0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
            "  0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
            "  0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
            "  0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
            "  0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
            "  0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
            "  0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
            "  0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.\n",
            "  0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
            "  0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.\n",
            "  0.]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# now create LSTM model\n",
        "from keras.layers import LSTM\n",
        "model = keras.models.Sequential()\n",
        "model.add(keras.layers.Embedding(50000, maxlen_Xtrain)) #limiting to 50 000 words\n",
        "model.add(keras.layers.LSTM(100)) # 100 memory units\n",
        "model.add(keras.layers.Dense(num_classes, activation='softmax')) # softmax activation for multi-class classification\n",
        "model.summary()\n",
        "# large model as we can see from the params"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uhl8VkGvW4Bz",
        "outputId": "655acc6e-b8e6-442e-a21c-25b6d4c0b14d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " embedding (Embedding)       (None, None, 232)         11600000  \n",
            "                                                                 \n",
            " lstm (LSTM)                 (None, 100)               133200    \n",
            "                                                                 \n",
            " dense (Dense)               (None, 25)                2525      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 11,735,725\n",
            "Trainable params: 11,735,725\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "sgd = keras.optimizers.Adam(learning_rate=0.001)\n",
        "model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])\n",
        "# cat crossentropy bc its multiclass, adam because that worked best in last model"
      ],
      "metadata": {
        "id": "g0nux53EYuCP"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "num_epochs = 20\n",
        "history = model.fit(padded_Xtrain, Y_train, epochs=num_epochs,\n",
        "                     validation_data=(padded_Xval, Y_val), batch_size=100)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ouAO3yD9ZPL8",
        "outputId": "ad9c2056-59f6-4821-d48e-0fefad011d8a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/20\n",
            "320/320 [==============================] - 29s 73ms/step - loss: 2.5821 - accuracy: 0.2220 - val_loss: 2.4883 - val_accuracy: 0.2360\n",
            "Epoch 2/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 2.4477 - accuracy: 0.2697 - val_loss: 2.2591 - val_accuracy: 0.3501\n",
            "Epoch 3/20\n",
            "320/320 [==============================] - 25s 79ms/step - loss: 2.2399 - accuracy: 0.3409 - val_loss: 2.0062 - val_accuracy: 0.4165\n",
            "Epoch 4/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 2.0209 - accuracy: 0.3956 - val_loss: 1.7708 - val_accuracy: 0.4741\n",
            "Epoch 5/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.8218 - accuracy: 0.4500 - val_loss: 1.6011 - val_accuracy: 0.5308\n",
            "Epoch 6/20\n",
            "320/320 [==============================] - 23s 70ms/step - loss: 1.6534 - accuracy: 0.4937 - val_loss: 1.4447 - val_accuracy: 0.5679\n",
            "Epoch 7/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.5156 - accuracy: 0.5308 - val_loss: 1.3210 - val_accuracy: 0.5988\n",
            "Epoch 8/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.4047 - accuracy: 0.5604 - val_loss: 1.2325 - val_accuracy: 0.6222\n",
            "Epoch 9/20\n",
            "320/320 [==============================] - 23s 70ms/step - loss: 1.3174 - accuracy: 0.5849 - val_loss: 1.1591 - val_accuracy: 0.6442\n",
            "Epoch 10/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.2461 - accuracy: 0.6038 - val_loss: 1.1079 - val_accuracy: 0.6608\n",
            "Epoch 11/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.1862 - accuracy: 0.6208 - val_loss: 1.0581 - val_accuracy: 0.6698\n",
            "Epoch 12/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.1352 - accuracy: 0.6379 - val_loss: 1.0035 - val_accuracy: 0.6845\n",
            "Epoch 13/20\n",
            "320/320 [==============================] - 23s 72ms/step - loss: 1.0883 - accuracy: 0.6497 - val_loss: 0.9659 - val_accuracy: 0.6984\n",
            "Epoch 14/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.0499 - accuracy: 0.6634 - val_loss: 0.9345 - val_accuracy: 0.7059\n",
            "Epoch 15/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.0121 - accuracy: 0.6754 - val_loss: 0.8988 - val_accuracy: 0.7191\n",
            "Epoch 16/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.9802 - accuracy: 0.6831 - val_loss: 0.8727 - val_accuracy: 0.7231\n",
            "Epoch 17/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.9454 - accuracy: 0.6963 - val_loss: 0.8485 - val_accuracy: 0.7340\n",
            "Epoch 18/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.9205 - accuracy: 0.7018 - val_loss: 0.8234 - val_accuracy: 0.7425\n",
            "Epoch 19/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.8932 - accuracy: 0.7127 - val_loss: 0.7993 - val_accuracy: 0.7498\n",
            "Epoch 20/20\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.8645 - accuracy: 0.7233 - val_loss: 0.7667 - val_accuracy: 0.7639\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_loss = history.history['loss']\n",
        "dev_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(dev_loss) + 1)\n",
        "\n",
        "plt.plot(epochs, train_loss, 'b', label='train')\n",
        "plt.plot(epochs, dev_loss, 'r', label='dev')\n",
        "plt.title('Loss curves')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Binary Crossentropy')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "dYIiVi2kaf0T",
        "outputId": "69176cd1-e277-4f27-8a67-0021bed806ce"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3hUZfbA8e8BQkc6FkBAFhVQCB0LCNiQInYQRayIurbV/Yndtbt21FVRUWygq6iIYhdBESQgvQgqLiAC0qu08/vj3DFDnCSTZO5Myvk8z32Y3DYnMc7Jfct5RVVxzjnnsiqV6gCcc84VTp4gnHPOxeQJwjnnXEyeIJxzzsXkCcI551xMniCcc87F5AnCOedcTJ4gXLEmIktE5LhUx+FcUeQJwrlCTETKpDoGV3J5gnAlkoiUE5HHROTXYHtMRMoFx2qJyFgRWS8ia0VkooiUCo7dICLLRWSTiCwUkWOzuX8FEXlYRH4RkQ0i8nWwr4uILMty7p9POSJyh4i8JSKvishG4CYR2SYiNaLObyUiv4tIWvD1hSIyX0TWicjHItIg2C8i8qiIrBKRjSIyW0QOC+UH6oolTxCupLoZ6AikAy2B9sAtwbHrgGVAbWBf4CZAReQQ4O9AO1WtApwILMnm/g8BbYAjgRrA/wF74oytD/AWUA14EPgWOD3qeH/gLVXdKSJ9gvhOC+KdCIwMzjsB6AwcDFQFzgLWxBmDc54gXIl1DnCnqq5S1dXAv4ABwbGdwP5AA1XdqaoT1YqW7QbKAc1EJE1Vl6jqj1lvHDxtXAhcrarLVXW3qk5S1T/ijO1bVX1XVfeo6jbgdeDs4N4C9Av2AQwG7lPV+aq6C7gXSA+eInYCVYBDAQnOWZG3H5MryTxBuJLqAOCXqK9/CfaB/dW+GPhERH4SkSEAqroYuAa4A1glIqNE5AD+qhZQHvhL8ojT0ixfvw0cISL7Y08Ee7AnBYAGwONBc9h6YC0gQF1V/QJ4EngqiHeYiOyTz5hcCeQJwpVUv2IfrhEHBvtQ1U2qep2qHgScDPwj0tegqq+r6tHBtQo8EOPevwPbgcYxjm0BKka+EJHSWNNQtL1KLKvqOuAToC/WvDRKM8swLwUuVdVqUVsFVZ0UXDtUVdsAzbCmpn/m9ENxLponCFcSpIlI+aitDNZOf4uI1BaRWsBtwKsAItJLRP4WNOdswJqW9ojIISLSLejM3g5sI0a/gqruAYYDj4jIASJSWkSOCK77ASgvIj2DTuZbsGar3LwOnAecQWbzEsAzwI0i0jyIvaqInBm8biciHYL32RLEHG8/iHOeIFyJ8CH2YR7Z7gDuBjKAWcBsYHqwD6AJ8BmwGesg/o+qfol9kN+PPSH8BtQBbszmPa8P7jsVa/Z5ACilqhuAy4HngeXYB/eybO4RbUwQ12+qOjOyU1XfCe49Khj1NAc4KTi8D/AcsA5rQluDNZ85FxfxBYOcc87F4k8QzjnnYvIE4ZxzLiZPEM4552LyBOGccy6mYlUIrFatWtqwYcNUh+Gcc0XGtGnTflfVrHNxgGKWIBo2bEhGRkaqw3DOuSJDRH7J7pg3MTnnnIvJE4RzzrmYPEE455yLqVj1QTjnXF7t3LmTZcuWsX379lSHEqry5ctTr1490tLS4r7GE4RzrkRbtmwZVapUoWHDhlh9xuJHVVmzZg3Lli2jUaNGcV/nTUzOuRJt+/bt1KxZs9gmBwARoWbNmnl+SvIE4Zwr8YpzcojIz/foCQK4+26YMSPVUTjnXOFS4hPEmjUwbBgccwx89VWqo3HOlTTr16/nP//5T56v69GjB+vXrw8hokwlPkHUrAnffAN168KJJ8K776Y6IudcSZJdgti1a1eO13344YdUq1YtrLAATxAA1K8PEydCejqcfjq88EKqI3LOlRRDhgzhxx9/JD09nXbt2tGpUydOPvlkmjVrBsApp5xCmzZtaN68OcOGDfvzuoYNG/L777+zZMkSmjZtyiWXXELz5s054YQT2LZtW0Ji82GugZo14fPPLUFcfDGsWgVDhkAJ6LtyzgWuuSbx/ZHp6fDYY9kfv//++5kzZw4zZsxg/Pjx9OzZkzlz5vw5HHX48OHUqFGDbdu20a5dO04//XRq1qy51z0WLVrEyJEjee655zjrrLN4++23Offccwscuz9BRKlUCcaMgf794aab4B//gD2+xLtzLonat2+/11yFoUOH0rJlSzp27MjSpUtZtGjRX65p1KgR6enpALRp04YlS5YkJBZ/gsiibFl45RWoXduy/urV8OKLkIfJh865Iiqnv/STpVKlSn++Hj9+PJ999hnffvstFStWpEuXLjHnMpQrV+7P16VLl05YE1NoTxAiUl9EvhSReSIyV0SujnFOFxHZICIzgu22qGPdRWShiCwWkSFhxRlLqVLw6KNw773w2mvQpw9s2ZLMCJxzJUWVKlXYtGlTzGMbNmygevXqVKxYkQULFjB58uSkxhbmE8Qu4DpVnS4iVYBpIvKpqs7Lct5EVe0VvUNESgNPAccDy4CpIjImxrWhEYEbb7QniUsvheOOgw8+gBo1khWBc64kqFmzJkcddRSHHXYYFSpUYN999/3zWPfu3XnmmWdo2rQphxxyCB07dkxqbKElCFVdAawIXm8SkflAXSCeD/n2wGJV/QlAREYBfeK8NqEuvtg6sM8+Gzp1go8/hnr1kh2Fc644e/3112PuL1euHOPGjYt5LNLPUKtWLebMmfPn/uuvvz5hcSWlk1pEGgKtgCkxDh8hIjNFZJyINA/21QWWRp2zLNgX696DRCRDRDJWr16dvwC/+AJ+yXZRJU49FT76CJYuhSOPhAUL8vc2zjlXlISeIESkMvA2cI2qbsxyeDrQQFVbAk8AeZ6mpqrDVLWtqratXTvmsqo5W7fOOhkuuwxUsz2tSxebab1jBxx9NHz3Xd7fyjnnipJQE4SIpGHJ4TVVHZ31uKpuVNXNwesPgTQRqQUsB+pHnVov2Jd41atbb/S4cZDNY15Eq1Y267pqVejWDT79NJSInHOuUAhzFJMALwDzVfWRbM7ZLzgPEWkfxLMGmAo0EZFGIlIW6AeMCStWLr8cOnaEq6+2ca05aNwYvv4a/vY36NkT3ngjtKiccy6lwnyCOAoYAHSLGsbaQ0QGi8jg4JwzgDkiMhMYCvRTswv4O/AxMB94U1XnhhZp6dLw/POwcaPNjsvF/vvD+PFwxBHWef3kk6FF5pxzKRPmKKavgRwLVajqk0DMj9egyenDEEKLrXlzG9d6551wzjnQvXuOp1erZh3XZ58NV14Ju3fbA4hzzhUXXmoj2k03QdOmNvFh8+ZcT69QAd56C045Bf75T5g+PQkxOueKvTvuuIOHHnoo1WF4gthLuXLW1LR0KdxyS1yXlCljl9SubTWctm4NOUbnnEsSTxBZHXmkdVoPHQpxTmuvWRNefhkWLoQEzlFxzpUg99xzDwcffDBHH300CxcuBODHH3+ke/futGnThk6dOrFgwQI2bNhAgwYN2BNUEt2yZQv169dn586dCY/Ji/XFct998N57No16+nSr4JeLY4+F666Dhx+GHj2gV69cL3HOFTapqPcNTJs2jVGjRjFjxgx27dpF69atadOmDYMGDeKZZ56hSZMmTJkyhcsvv5wvvviC9PR0vvrqK7p27crYsWM58cQTSQuhoqg/QcRSpQo8/TTMnQsPPBD3ZffcAy1bwoUXwsqVIcbnnCtWJk6cyKmnnkrFihXZZ599OPnkk9m+fTuTJk3izDPPJD09nUsvvZQVK1YA0LdvX94IxtiPGjWKvn37hhKXP0Fkp1cv6NcP7r4bzjjDOq9zUa6cVX9t29aSxNixvuCQc0VKYaj3HdizZw/VqlVjRownmpNPPpmbbrqJtWvXMm3aNLp16xZKDP4EkZPHH4fKla2pKc6Vg5o3hwcfhA8/hHysQ+6cK4E6d+7Mu+++y7Zt29i0aRPvv/8+FStWpFGjRvz3v/8FQFWZOXMmAJUrV6Zdu3ZcffXV9OrVi9KlS4cSlyeInNSpYwtDTJoEzzwT92VXXAEnnWQd1vOSXn/WOVfUtG7dmr59+9KyZUtOOukk2rVrB8Brr73GCy+8QMuWLWnevDnvvffen9f07duXV199NbTmJQDRHArUFTVt27bVjIyMxN5U1SbNTZpkn/b16+d+DfDbb3D44VYafPJka35yzhU+8+fPp2kcTcjFQazvVUSmqWrbWOf7E0RuROzpYc8eG/4aZ0Ldbz8YPtwGRNx6a8gxOudcCDxBxKNRI+usHjsW3nwz7st694bBg+Ghh2zJCeecK0o8QcTrqqugXTsrvLRmTdyXPfwwHHwwnHcerF0bYnzOuXwrTk3t2cnP9+gJIl6Riq/r1tmMuDhVrGhDX1eutKeJEvB76FyRUr58edasWVOsk4SqsmbNGsqXL5+n63weRF60aAE33GAz4s45B44/Pq7L2rSBu+6yYrE9e8LAgSHH6ZyLW7169Vi2bBn5XrK4iChfvjz16tXL0zU+iimvtm+36dI7dsCcOVCpUlyX7d5t5TimTbOO68aNww3TOefi4aOYEql8eXjuOViyBG67Le7LSpe2gn6lS8OAAbBrV3ghOudcIniCyI/Ona1D4bHHYOrUuC878EAbMfvtt9ZK5ZxzhZkniPy6/36b7HDxxZCHMrv9+sG551qfxLffhhifc84VUGgJQkTqi8iXIjJPROaKyF8W5BSRc0RklojMFpFJItIy6tiSYP8MEQm5YyEfqla1YkuzZlnxpTx48kmbkH3uubBpU0jxOedcAYX5BLELuE5VmwEdgStEpFmWc34GjlHVw4G7gGFZjndV1fTsOlBSrk8fOPNMW8c6WOAjHlWrwquvWjfGVVeFF55zzhVEaAlCVVeo6vTg9SZgPlA3yzmTVHVd8OVkIG9jsAqDoUNtceorrsjTJIejjrIlsF96yda1ds65wiYpfRAi0hBoBUzJ4bSLgHFRXyvwiYhME5FBOdx7kIhkiEhGSsYx77efPUF8/rmtQpcHt90G7dvDoEGwbFlI8TnnXD6FPg9CRCoDXwH3qOrobM7pCvwHOFpV1wT76qrqchGpA3wKXKmqE3J6r6TMg4hl1y5bVnDrVqv4mofZiosWQatWthT2xx/7AkPOueRK2TwIEUkD3gZeyyE5tACeB/pEkgOAqi4P/l0FvAO0DzPWAilTxhYX+vlneOSRPF3apIkV8/v0U5te4ZxzhUWYo5gEeAGYr6oxPzVF5EBgNDBAVX+I2l9JRKpEXgMnAHPCijUhjj0WTj0V7r0Xli/P06WXXgrHHWclnpYsCSc855zLqzCfII4CBgDdgqGqM0Skh4gMFpHBwTm3ATWB/2QZzrov8LWIzAS+Az5Q1Y9CjDUxHn7YmptuuCFPl4lYHUARuOiiuFc3dc65UHktpkS75RabJv3NN9axkAfPPw+XXAJPPWVrEznnXNhy6oPwBJFomzfDIYfA/vvDd99Bqfgf0lRtLeuJE2H2bDjooBDjdM45vFhfclWuDP/+t5VtfemlPF0aaWpKS4MLLvCmJudcanmCCEP//ta8dOONsGFDni6tV89qAE6YYCU5nHMuVTxBhEHEZlivXm1V+fJo4EBbWGjIEJsn4ZxzqeAJIixt2sCFF9r8iDzUaQLLL8OGQbly1tS0e3dIMTrnXA48QYTp3nttUeprr83zpQccAE88YYOhHn88hNiccy4XniDCVKcO3H47jBsHH3yQ58vPOccKxt58MyxYEEJ8zjmXA08QYfv7323Y67XX2jrWeSBiK9BVrAjnn+9NTc655Mo1QYjIwyLSPBnBFEtly8Kjj1pvcz7aivbbzybOTZliE7Wdcy5Z4nmCmA8ME5EpQZmMqmEHVeycdJINS7rrLvjttzxf3rcvnH463HqrFYt1zrlkyDVBqOrzqnoUcB7QEJglIq8HJbpdvB59FLZvt1WC8kjEVjfdZx8bArtrVwjxOedcFnH1QYhIaeDQYPsdmAn8Q0RGhRhb8dKkCVxzDbz4IkydmufL69SxJJGRYRO1nXMubLnWYhKRR4HewOfAC6r6XdSxhap6SLghxq9Q1GLKycaNcPDB0KiRjV/NQ52miH79YPRoq+Rx+OEhxOicK1EKWotpFtBSVS+NTg6BwruIT2G0zz5w//0weTK89lq+bvHkk1C9ujU17dyZ4Piccy5KPAniJeBEEXkkGNF0auSAquat0JCD886zhahvuAE2bcrz5bVq2dDX77+H++4LIT7nnAvEkyCeAgYDs7FV3S4VkadCjao4K1XK6jStWGEzrfPh1FNtEt1dd8GMGQmOzznnAvH0QSwAmmpwooiUAuaqatMkxJcnhb4PItrAgTBqFMydC3/7W54vX7sWmje3zuupU226hXPO5VVB+yAWAwdGfV0/2OcK4r77bOGH66/P1+U1alhBv1mz4O67Exybc84RX4KoAswXkfEiMh6YB+wjImNEZEx2F4lIfRH5UkTmichcEbk6xjkiIkNFZLGIzBKR1lHHBorIomAbmI/vrXA74ABbnvS99+DTT/N1i969rUvj3nttVJNzziVSPE1Mx+R0XFW/yua6/YH9VXW6iFQBpgGnqOq8qHN6AFcCPYAOwOOq2kFEagAZQFtAg2vbqOq6nGIpUk1MAH/8Ye1EZcvCzJn2RJFH69bZcNdKlSxJVK4cQpzOuWKrQE1MQQJYgD1JVAHmq+pXkS2H61ao6vTg9SasZEfdLKf1AV5WMxmoFiSWE4FPVXVtkBQ+Bbrn+p0WNeXKwSOPwPz5VnApH6pXh1dfhcWL4YorEhyfc65Ei6dY31nAd8CZwFnAFBE5Iy9vIiINgVbAlCyH6gJLo75eFuzLbn/x07u31Wq69VZYtixft+jSBW67DV5+GUaMSGx4zrmSK54+iJuBdqo6UFXPwybH3RrvG4hIZeBt4BpV3Zi/MHO8/yARyRCRjNWrVyf69uETsaeH3bvhqqvyfZtbbrFEcfnlvnaEcy4x4kkQpVR1VdTXa+K8DhFJw5LDa6o6OsYpy7FRURH1gn3Z7f8LVR2mqm1VtW3t2rXjCavwadTIFhZ65x3rtM6H0qVtcnbFilb9ddu2BMfonCtx4vmg/0hEPhaR80XkfOAD4MPcLhIRAV7A+iweyea0McB5wWimjsAGVV0BfAycICLVRaQ6cEKwr/j6xz/gsMNsgaF8zLAGGxj18ss29PW66xIcn3OuxMkxQQQf8kOBZ4EWwTZMVW+I495HAQOAbiIyI9h6BGtKDA7O+RD4CZtX8RxwOYCqrgXuAqYG253BvuIrLc0mNixbZk8T+XTSSTa14umn4e23Exifc67EiWeY62xVLRJ1Q4vcMNdYLrvMEsXUqdC6de7nx7BjB3TqBAsXWs2mRo0SHKNzrtgo6Ezq6SLSLsExuezcdx/Urg2DBuV7EeqyZa2KB1h58Dwuhe2cc0B8CaID8K2I/BjMdp4tIrPCDqzEqlbN1q6eNi3fcyPAnhpeeAG++w5uvjmB8TnnSox4mpgaxNqvqr+EElEBFIsmJgBV6NEDvv7aJtHVq5fvW11+ufVHfPih9U8451y0gjYx3a2qv0RvgJeHC1NkEeoCzo0Am6jdooXVbFoec6Cwc87FFk+CaB79RbA+dZtwwnF/SsDcCIDy5eGNN2DrVjj33Hx3azjnSqBsE4SI3Cgim4AWIrIx2DYBq4D8f2K5+CVgbgTAoYfaA8n48V4a3DkXv2wThKrep6pVgAdVdZ9gq6KqNVX1xiTGWHIlaG4E2PpE550Hd95picI553KTayc1gIjUBRoAZSL7VHVCiHHlS7HppM4qAXMjADZvhrZtYeNGqy5eVCuTOOcSJ6dO6nhGMd0P9MMWCoq0YKuqnpzQKBOg2CaI9eutnahePZgyxQov5dPMmdChA3TrBmPH2hLZzrmSq6CjmE4FDlHVHqraO9gKXXIo1hI0NwKgZUsb2TRunP3rnHPZiSdB/ATkfakzl1hnnQXdu9ust3yuGxFx2WVw+ulw4432QOKcc7HEkyC2AjNE5Nlg/eihIjI07MBcFgmcGyECzz9vLVb9+lkLlnPOZRVPghiDVVadhK0NHdlcsiVobgRYq9WoUfYwcvHFNnnbOeeixTuKqQJwoKouDD+k/Cu2ndTRdu6ENm1g3TqYNw+qVCnQ7R58EP7v/2DoULjyygTF6JwrMgrUSS0ivYEZwEfB1+kiMiaxIbq4paXBs89a3Yzbbivw7a67zpbFvuYaq9fknHMR8TQx3YGtQ70eQFVnAAeFGJPLzRFHwODB9mf/tIK19pUqBa+/Dunp1g9ewNs554qReBLETlXdkGXfnjCCcXlw771Qp46tG7FrV4FuVbkyfPAB1KoFvXrBL4WuTq9zLhXiSRBzRaQ/UFpEmojIE1iHtUulyNyI6dMLPDcCYL/9rIlp2zYrC75uXQJidM4VafEkiCuxiq5/ACOBjcA1YQbl4nTmmfZpfsstsHRpgW/XrBm8+y4sXgynnQZ//JGAGJ1zRVauCUJVt6rqzaraDltd7gFV3Z7bdSIyXERWicicbI7/U0RmBNscEdktIjWCY0uCletmiEgxH5ZUACL29LB7t9XyTsAnepcu8NJLVtDvoot8+KtzJVk8o5heF5F9RKQSMBuYJyL/jOPeLwHdszuoqg+qarqqpgM3Al+p6tqoU7oGx2MOv3KBRo1s1tuECQmb0NC/v3VxvPYa3HprAmJ0zhVJ8TQxNVPVjcApwDigETAgt4uCaq9rczsvcDbWfOXyo39/W+jh1VfhjjsScsshQ+CSS+Cee+C55xJyS+dcERNPgkgTkTQsQYxR1Z1AwhoeRKQi9qTxdtRuBT4RkWkiMiiX6weJSIaIZKxevTpRYRU9N90EF15oCz689FKBbxep7NG9u9VuGjeu4CE654qWeBLEs8ASoBIwQUQaYB3VidIb+CZL89LRqtoaOAm4QkQ6Z3exqg5T1baq2rZ2SV7gQASeeQaOO87+9P/88wLfskwZePNNW9P6rLPg++8TEKdzrsiIp5N6qKrWDcp9q6r+AnRNYAz9yNK8pKrLg39XAe9gE/VcbtLS4K234JBDrFzr3LkFvmWVKrZuRPXq0LMn/O9/CYjTOVckxNNJfXXQSS0i8oKITAe6JeLNRaQqcAxRa1yLSCURqRJ5DZwAxBwJ5WKoWtUmNFSoAD16wG+/FfiWBxxgTUxbt9otvfqrcyVDPE1MFwad1CcA1bEO6vtzu0hERgLfAoeIyDIRuUhEBovI4KjTTgU+UdUtUfv2Bb4WkZnAd8AHqvpRnN+PAzjwQJsavWaNTY3esiX3a3LRvDmMHg0//GAPJzt2JCBO51yhFs+So7NUtYWIPA6MV9V3ROR7VW2VnBDjVyKquebF2LHQp4+1Db3zToGWKo145RU47zwYMABGjLCuD+dc0VXQJUenicgnQA/g46D5x2sxFQW9esETT8D778O11ybklgMGwF13WaK4/faE3NI5V0iVieOci4B04CdV3SoiNYELwg3LJczll8OPP9oC1I0bw9VXF/iWN98MS5ZYomjY0EbXOueKn1wThKruEZF6QH+x9oSvVPX90CNzifPgg/aJfu210KABnHJKgW4nAk8/bavRDRoEdevCiScmJlTnXOERzyim+4GrgXnBdpWI3Bt2YC6BSpWyNqH27W3W9XffFfiWaWk2R+Kww+CMM2DGjATE6ZwrVOLpg+gBHK+qw1V1ODbruVe4YbmEq1gRxoyxut69e8PPPxf4lvvsY4OlqlWDY4+FqVMTEKdzrtCIJ0EAVIt6XTWMQFwS1KljcyR27rQJDQlY9KFuXfjqK5t+0a2bVYF1zhUP8SSIe4HvReQlERkBTAPuCTcsF5pDD7Uhrz/+aIs+JGBCw0EHwddf2/SL7t1t0JRzrujLMUGISClsSGtHYDRWUO8IVX0jCbG5sBxzDLz4ov25n6AS4QccYBXHDz8cTj3V1rl2zhVtOY5iCkYw/Z+qvgmMSVJMLhnOOcf6IW691R4BElAmvGZNqxF48sm2ftGGDVYJ1jlXNMXTxPSZiFwvIvVFpEZkCz0yF76bb4YLLoB//cueKBJgn32sblPPnjYF4/5ci7I45wqreCbK9Q3+vSJqnwIHJT4cl1Qi8OyzNqHh4outp/m00wp82woVrG7T+efDjTdacb/77vOyHM4VNfFMlGuUjEBciqSlWaf18cdDv35Wv+mEExJy21desSeKBx6w5qannrIpGc65oiHb/11F5FwR+cvSoiIyQET6hxuWS6pKlWz4a7NmNsv6m28ScttSpWxVuiFDbC2jAQNshK1zrmjI6e+5K7HFerIaDVwXTjguZapVg08+gfr1bY5EgpaPE7Hmpfvus5FNp50G27Yl5NbOuZDllCDSVHVz1p3B2g1p4YXkUqZOHfjsM0sWJ5wACxYk7NZDhlj9pg8+sPyzaVPCbu2cC0lOCaJCsKLbXoJy32XDC8mlVP368Omn1j503HFW5C9BBg+GV1+FiRNt1vXvvyfs1s65EOSUIF4A3hKRBpEdItIQGBUcc8XVwQdbktiyxZLEihUJu3X//tYnPnu2zddbvjxht3bOJVi2CUJVH8LWip4gImtEZA3wFTBWVR9MVoAuRVq0sAkNv/1mzU1r1ybs1r17w0cfwf/+B506wU8/JezWzrkEynHQoao+o6oNgIZAQ1VtoKpPx3NjERkuIqtEZE42x7uIyAYRmRFst0Ud6y4iC0VksYgMycP34xKpY0d47z1biPqkkxLacdClC3zxhQ1/Pfpoe6JwzhUucY1KV9VNqprXT4eXsNLgOZmoqunBdieAiJQGngJOApoBZ4tIszy+t0uUY4+1hR+mTbP1rbdvT9it27Wz+k0iloteeSVht3bOJUBo05ZUdQKQn3aJ9sBiVf1JVXdgfR59Ehqcy5s+fWDECCvud9ZZCZ3M0Ly5rSPRti2cd55N6PZhsM4VDqme13qEiMwUkXEi0jzYVxdYGnXOsmBfTCIySEQyRCRj9erVYcZasp1zjk2Ffv99GDgQdu9O2K0POMCK/N18M7zwAnToAAsXJuz2zrl8imfJ0WkicoWIVE/we08HGqhqS+AJ4N383ERVh1i1YsAAABu7SURBVKlqW1VtW7t27YQG6LK47DKb8TZypFXiS0CZ8IgyZeDuu63zesUKaNPGS4Y7l2rxPEH0BQ4AporIKBE5UaTgZddUdWNkIp6qfgikiUgtYDlQP+rUesE+VxgMGWLbsGFwww0JTRIAJ55ok7hbtbKHlsGDE9rt4ZzLg1wThKouVtWbgYOB14HhwC8i8q+ClP0Wkf0iiUZE2gexrAGmAk1EpJGIlAX64WtRFC733mtPEA8+aK8TrF49+PJLy0PPPmsd2IsWJfxtnHO5iKsPQkRaAA8DD2Kryp0JbAS+yOGakcC3wCEiskxELhKRwSIyODjlDGCOiMwEhgL91OwC/g58DMwH3lTVufn79lwoROCJJ2xVoFtusdcJVqaMtWZ98AEsXWpNTm++mfC3cc7lQDSXJgIRmQasx2ZPv62qf0QdG62qBV9AIEHatm2rGRkZqQ6j5Ni1C844w+ZKDBsGl1wSytssXQp9+8K339qDy8MPQ/nyobyVcyWOiExT1baxjsWzJvXbqnqsqr4enRwAClNycClQpgyMGgXdu8OgQdYnsWdPwt+mfn346iu4/norH37UUfDjjwl/G+dcFrnNpN4DeBJw2StfHsaMsd7kf//b6nlv/ksR4AJLS7MujzFjbCnt1q3h7bcT/jbOuSi+JrUruLQ0+9P+8cdtnkSnTraMaQh697ZRToceaq1bV10Ff/yR+3XOubyLd5jrFcAEYFqweUO/25uIfVqPHWvtP+3a2RTpEDRoYCXDr73W+sePPtqbnJwLQzzDXBvF2A5KRnCuCDrpJOtNLl8eOncObehR2bLwyCNWOnzRIjjsMLjzTi/T4VwixTvM9TAROUtEzotsYQfmirDmzeG772xsat++NkU6wRPqIk45BebMsXJRt99ubz1mTGhv51yJEk+pjduxUhhPAF2BfwMnhxyXK+pq17YCSwMGwK232pyJkKZE16tng6m++AIqVLBk0aOHVSl3zuVfPE8QZwDHAr+p6gVAS6BqqFG54qFcOasCe889VlipWzdYuTK0t+vaFWbMgEcfhUmTrNnpxhtDGVTlXIkQT4LYFgx33SUi+wCr2LtWknPZE4GbboK33rJP7/btQ10dKC0NrrnGqsH27w/3328jnt54w5udnMureBJEhohUA57DRjBNx0poOBe/00+3oUe7dsGRR1oNjRDttx+89BJ88w3UqQP9+tkDzJyY6xs652KJZxTT5aq6XlWfAY4HBgZNTc7lTZs21nl98MFw8snWFhTyn/VHHmmjbZ9+GmbOhPR0Gx67YUOob+tcsRDvKKa6InIkcCBQTUQ6hxuWK7bq1rV1Rk85Bf7xD5uBncAV6mIpXdre5ocfbMW6xx+3HDViRCiVQZwrNuIZxfQA8A1wC/DPYLs+5LhccVapEvz3v9Y3MWyY1XJam5/VafOmVi145hl7omjUCM4/3ybZTZ8e+ls7VyTF8wRxCnCIqvZQ1d7B5sNcXcGUKmWjm15+Gb7+2iYwvP56UnqS27SxUU4vvgiLF9t62BdeaK+dc5niSRA/AWlhB+JKqAED7NO6Xj1bQq5rV5gb/vIfpUrZE8QPP9iop5Ej4ZBDbLrGvHmhv71zRUI8CWIrMENEnhWRoZEt7MBcCdKmDUyebMvHzZ4NLVtabe9Nm0J/62rVrGTHzz9bl8i779rDzBlnWFFA50qyeBLEGOAuYBKZxfqmhRmUK4FKl7Y1JRYutPaeRx6xCQyjRiWl2Wm//ayc+JIltkjep59aSfFevSx3OVcS5bqiXFHiK8oVI1OmwBVXwLRp1uz05JPQrFnS3n79enjqKRuJu2YNHHusJY5jjrG5f84VF/laUU5E3gz+nS0is7JucbzpcBFZJSIxpyaJyDnBvWaLyCQRaRl1bEmwf4aI+Cd+SdShgyWJp5+2GdgtW8I//5mUZiewpqebb7Yniocesgl2XbtagdqPP/ZZ2a5kyPYJQkT2V9UVItIg1nFV/SXHG9tcic3Ay6p6WIzjRwLzVXWdiJwE3KGqHYJjS4C2qvp7Xr4Zf4IoplavtqJKL7xg8ygefhjOOiupf8pv2wbDh8MDD9ga2W3b2hNF797W4e1cUZWvJwhVXRH8+0tkA7YA/8stOQTXTQCyHdyuqpNUdV3w5WSgXm73dCVU7drw/PO2zkSkbsZxx8H8+UkLoUIFa/FavNhCWbvW5vqlp1udp927kxaKc0mTUxNTRxEZLyKjRaRV0FQ0B1gpIt0THMdFwLiorxX4RESmicignC4UkUEikiEiGatXr05wWK5Q6djRZrk99ZTNbmvRAm64IanlWsuWhYsusr70V16xSeD9+kGTJvZ0sWpV0kJxLnyqGnPDlhU9ATgTWAd0DPYfCnyf3XVZ7tEQmJPLOV2B+UDNqH11g3/rADOBzvG8X5s2bdSVECtXql5wgSqo1qun+sorqrt2JT2M3btV33pLtUsXCyUtTbVfP9Xx41X37El6OM7lGZCh2Xym5tR6WkZVP1HV/2JrQUwOEsqCRCUnEWkBPA/0UdU1kf2qujz4dxXwDtA+Ue/piok6daxTYNIka4IaMMAmMLz2WlLbe0qVskK1X35pE+wuvxw++gi6dLFBV48/DuvW5Xob5wqlnBJEdBmzrCv9FngMh4gcCIwGBqjqD1H7K4lIlchr7CnGizS72I44AjIybL2JsmVtKnQKEgVA06bw2GOwfLmV8aha1WZpH3AAXHCBFbL10U+uKMlpFNNurFNagArYjGqCr8urao7lN0RkJNAFqAWsBG4nKNmhqs+IyPPA6UCkw3uXqrYVkYOwpwaAMsDrqnpPPN+Mj2Iq4fbsgXfegX/9y2ZkH3ywLXfarx+UKZOSkL7/3goEvvYabNkCrVpZZdn+/aFy5ZSE5NxechrF5BPlXPGzZ4/VzPjXv2DWrEKRKDZutCTx9NOWu6pUsYedwYOtr925VMnXMFfniqxSpeC00+zP97ffhvLlM/soXn3VVrVLsn32gcsus0WLJk2yIbLDh9v8vyOPtNfr1yc9LOdy5AnCFV/RiWL06MxE0ayZjVFNQaIQsW6Tl1+2voqHH4bff7ehs/vua4njjTdg69bc7+Vc2DxBuOKvVCk49dTMRFGxIpx3XkoTBUDNmlZBduFCqypy+eXWkd2vnw3SOuccGDsWduxISXjOeYJwJUgkUUyfbp3ZkUTRtCn85z/2p3wKiED79lYYcOlS+OIL68QeN85Keey3H1xyie33GdsumbyT2pVce/bAmDFw5532dFG6NBx/vH06n3KK9SSn0I4d8MkntpjRe+/ZKKj997cyVGefbUnFK8u6gvJRTM7lRNVGO40caetP/PKL9Vf07GmfxD172tcptHWrNTeNHAkffmjJo1Eja446+2w4/PCUhueKME8QzsVL1YoCjhwJb75pxZWqVLGmqbPPtiKBKRoqG7F+vbWQjRwJn39uD0JNm1p//Gmn2VwLf7Jw8fIE4Vx+7NplNTRGjrTO7Q0boFYtOPNMSxZHHZXyWt8rV8J//2ujeSdMsGTRoIHls9NOsyG0pUunNERXyHmCcK6g/vjDeo1HjoT337cFIurXh759LVkUgj/bV6+20EaPtiVTd+yw0VB9+liy6NbNqpE4F80ThHOJtHmz9RqPHGnLy+3aZbO1zzrLEsZhf1kfK+k2brR8Nnq09Vls3my1oXr2tGTRvTtUqpTqKF1h4AnCubCsXWvtO2+8Yc1Re/bY/Iq+fS1hHHpoqiNk+3b47DNLFmPG2Brb5cvDiSdasujdG6pXT3WULlU8QTiXDCtXZiaLiROtw7tFi8xk8be/pTpCdu2y0N55xxLG8uXW53700XDssba1bQtpOZbidMWJJwjnku3XX60E+RtvWPElgNatM5NFw4YpDQ/sYScjw5LFRx/BjBm2v0oV6NzZkkW3bjaE1tfdLr48QTiXSv/7nw01evNNq6UBNsutb18bEVW/fmrjC/z+O4wfb0NnP/8cFi2y/bVqWaLo1s2SRuPGKe+PdwnkCcK5wuLnny1RvPmmlfwAGy579tk2661mzdTGFyVS9iOSMH791fYfeGDm00W3brYgkiu6PEE4VxgtWmSJ4o03bJGItDTo1QsGDoQePQpVR4Aq/PCDJYovvrD++LVr7VjTppnJoksXqFEjpaG6PPIE4VxhN3MmjBhhqwqtWmXtOv37W7IoBHMsstqzx/osIk8YEyZYORARSE+Hrl0tYXTqZGthuMLLE4RzRcXOnTa3YsQIG5O6Y4fNqxg40Op/779/qiOMaccOmDo18+li0iSbW1i6NLRrZ8mia1eb2V2xYqqjddFSliBEZDjQC1ilqn+ZPSQiAjwO9MDWvD5fVacHxwYCtwSn3q2qI3J7P08QrlhZu9aan0aMsAUjSpWyyQsDB9r06BQXEMzJtm1W0iqSML77zobYli1rCyZFmqTat/fZ3amWygTRGdgMvJxNgugBXIkliA7A46raQURqABlAW0CBaUAbVV2X0/t5gnDF1sKFliheeQWWLbNp0X37WrI44ohC1wSV1aZN8PXXmQlj+nTr16hY0eZgHHMMdOxoTxsprrJe4qS0iUlEGgJjs0kQzwLjVXVk8PVCoEtkU9VLY52XHU8QrtjbvdvGoo4YYZPytm6FJk1suOyxx1qyqFAh1VHmat06+OorSxaffw5z59p+EZuI3rEjdOhg/zZr5gUHw1SYE8RY4H5V/Tr4+nPgBixBlFfVu4P9twLbVPWhnN7LE4QrUTZtsiQxYoRNj969G8qVsyTRtattHToUiTacdeusD2PyZNumTMkcJVW5sj1ZRBJGhw62yp5LjJwSRGoL2yeAiAwCBgEceOCBKY7GuSSqUgXOP9+2jRv3bsO54w64/XZrwznqqMxe4jZtUr6eRSzVq8MJJ9gG1vy0eLElikjCeOihzOXDGzSwRBFJGq1bF+oumSIr1U8Q3sTkXBjWrrWxp5GEMWeO7Y/U0YgkjJYti0wdjW3bbGXYSMKYPNkmqYM9JLVta8NqO3Wy0VJegDA+hbmJqSfwdzI7qYeqavugk3oa0Do4dTrWSb02p/fyBOFcNlatsr6LL7+0pPHDD7a/enWb3da5s/053qpVkfpTfMUKSxaTJlkrW0aGPWWI2OjgTp2sE7xTJ6hXL9XRFk6pHMU0EnsaqAWsBG4H0gBU9ZlgmOuTQHdsmOsFqpoRXHshcFNwq3tU9cXc3s8ThHNxWr7ckkUkYSxZYvvLlLGnikj7Tfv2ttZFEXnK2LrVhtROnGgtbpMm2VoYYPURoxPGoYcW+sFfSeET5ZxzOfv1V/tknTLFtqlTMz9Zq1a1RNG+fWbiqFMntfHGadcum6T+9deWNCZOtIcpsLJXkWTRsSM0bw7VqqU23lTwBOGcy5vdu2HBgsyE8d13Vi9q9247Ht1L3KGD9RIXgeG1kc7vyBPGxIn2dcQBB1iiOOww+7d5cxtmW5zLhXiCcM4V3JYtNsMtkjCmTMnsJU5LsyeMzp1tO/LIIvOpumKFfVtz59o2Zw7Mn2+d4hH16++dNCKJozgs2+oJwjkXjt9+y+wlnjAhs5e4VCnr8I4kjKOPtgKERcTu3dYtE0kYkeSxYIHVmIpo1CjziSM93bpvmjQpWhP7PEE455JjyxYbfzphgm2TJ9ui2GCfpJGE0blzkVxIYtcu+PHHzIQRSSALF2bO0ahQwVbhS0/PTBotWtiEv8LIE4RzLjX++MOeKiIJ4+uvMzu/GzfOTBZHHGFrdhelP72j/PEHzJtnHeIzZmT+u369HRexbzeSMCL/1quX+pFUniCcc4VDZFhRJGFMmJBZU6N8eXvKOPxw+5M78m8RGTGVlaqtyhedMGbMgJ9+yjynRg1LFK1aWT9/69Y2qjiZedIThHOucNqzx/70zsiwUVKzZtm/K1dmnlOnzl+TRrNmRWLUVCwbN9q3GZ00Zs/O7NuoWNGSRiRhtG5t325YJbU8QTjnipZVq+xTMzppzJmT2Z9RqpQ1SUWSxuGH26do48aFstZUbnbutA7w6dNt+/572yKtcWXL2rcYnTQOPzwxOdIThHOu6Nu923qIIwkjkjx++snac8A+SQ8+2JJFZGve3JJJEahqG23PHpujEUkakW1dsCpO6dL27UUSxpVX5q8/wxOEc6742rzZJi7Mm7f39vPPmYmjTBkbfxqdOJo1s2RShGpPqcIvv2Q+ZUyfDtOm2bcQqZaSV54gnHMlz9atNv40a+JYvNj+PAdrqmrc2JLFYYdlbgcfXKSeODZssIoo+VGs14NwzrmYKla04UGtWu29f/t2WLQoM2FEJjSMHZtZSqRMGUsSkenTkcTRuHGhHIqb3+SQG08QzrmSpXz5zI7taH/8YU8ckanTc+bY6Ko338w8p1w5aNo0M2FEkseBBxaZird54QnCOefAPvxbtLAt2pYt1scxZ05m8hg/Hl59NfOcSpUscWTt42jYsFA+ccTLE4RzzuWkUiVbrq5tlmb69eutiWr27Mzmqs8+g5dfzjynfHlbeCJr8mjc2AocFnKeIJxzLj+qVbOqtUceuff+DRv+Oqpq0iQYGbViclra3sNxDz/cFqWoWze530MuPEE451wiVa1qH/YdO+69f/Pmv46q+v57ePvtzFFV9etbXarIlp5uTV8p4gnCOeeSoXJlaNPGtmjbttmEv8mT4dtvbYt0jJcrZ7PgIgmjY8ekLq7t8yCcc66w+fVXSxSRpJGRkVmsqV69vRNG69YFespI2UQ5EekOPA6UBp5X1fuzHH8U6Bp8WRGoo6rVgmO7gdnBsf+p6sm5vZ8nCOdcsbRjh1X1i37K+OUXO1a2rC37On58vobapmSinIiUBp4CjgeWAVNFZIyqzouco6rXRp1/JRA9o2WbqqaHFZ9zzhUZZcvakq7t28NVV9m+FSsyE8a6daHMwwizD6I9sFhVfwIQkVFAH2BeNuefDdweYjzOOVd87L8/nHqqbSEJc+pfXWBp1NfLgn1/ISINgEbAF1G7y4tIhohMFpFTsnsTERkUnJexevXqRMTtnHOOcBNEXvQD3lLV3VH7GgTtYv2Bx0SkcawLVXWYqrZV1ba1a9dORqzOOVcihJkglgP1o76uF+yLpR8wMnqHqi4P/v0JGM/e/RPOOedCFmaCmAo0EZFGIlIWSwJjsp4kIocC1YFvo/ZVF5FywetawFFk33fhnHMuBKF1UqvqLhH5O/AxNsx1uKrOFZE7gQxVjSSLfsAo3Xu8bVPgWRHZgyWx+6NHPznnnAufT5RzzrkSLKd5EIWlk9o551wh4wnCOedcTMWqiUlEVgO/pDqObNQCfk91EDnw+ArG4ysYj69gChJfA1WNOUegWCWIwkxEMrJr5ysMPL6C8fgKxuMrmLDi8yYm55xzMXmCcM45F5MniOQZluoAcuHxFYzHVzAeX8GEEp/3QTjnnIvJnyCcc87F5AnCOedcTJ4gEkhE6ovIlyIyT0TmisjVMc7pIiIbRGRGsN2W5BiXiMjs4L3/UpdEzFARWSwis0SkdRJjOyTq5zJDRDaKyDVZzknqz09EhovIKhGZE7Wvhoh8KiKLgn+rZ3PtwOCcRSIyMInxPSgiC4L/fu+ISLVsrs3xdyHE+O4QkeVR/w17ZHNtdxFZGPwuDklifG9ExbZERGZkc20yfn4xP1OS9juoqr4laAP2B1oHr6sAPwDNspzTBRibwhiXALVyON4DGAcI0BGYkqI4SwO/YZN4UvbzAzoDrYE5Ufv+DQwJXg8BHohxXQ3gp+Df6sHr6kmK7wSgTPD6gVjxxfO7EGJ8dwDXx/Hf/0fgIKAsMDPr/0thxZfl+MPAbSn8+cX8TEnW76A/QSSQqq5Q1enB603AfLJZRa8Q6wO8rGYyUE1E9k9BHMcCP6pqSmfGq+oEYG2W3X2AEcHrEUCsFQ9PBD5V1bWqug74FOiejPhU9RNV3RV8ORlbiyUlsvn5xePPJYtVdQcQWbI4oXKKT0QEOIssa9UkUw6fKUn5HfQEERIRaYgtcjQlxuEjRGSmiIwTkeZJDQwU+EREponIoBjH414qNmR/WUQqSip/fgD7quqK4PVvwL4xziksP8cLsSfCWHL7XQjT34MmsOHZNI8Uhp9fJ2Clqi7K5nhSf35ZPlOS8jvoCSIEIlIZeBu4RlU3Zjk8HWs2aQk8Abyb5PCOVtXWwEnAFSLSOcnvnyuxBaZOBv4b43Cqf357UXuWL5RjxUXkZmAX8Fo2p6Tqd+FpoDGQDqzAmnEKo7PJ+ekhaT+/nD5Twvwd9ASRYCKShv2HfE1VR2c9rqobVXVz8PpDIE1s1byk0MylXFcB72CP8tHyslRsWE4CpqvqyqwHUv3zC6yMNLsF/66KcU5Kf44icj7QCzgn+AD5izh+F0KhqitVdbeq7gGey+Z9U/3zKwOcBryR3TnJ+vll85mSlN9BTxAJFLRZvgDMV9VHsjlnv+A8RKQ99t9gTZLiqyQiVSKvsc7MOVlOGwOcF4xm6ghsiHqUTZZs/3JL5c8vyhggMiJkIPBejHM+Bk4QWz63Ovaz/jgZwYlId+D/gJNVdWs258TzuxBWfNF9Wqdm875xLVkcouOABaq6LNbBZP38cvhMSc7vYJg98CVtA47GHvVmATOCrQcwGBgcnPN3YC42KmMycGQS4zsoeN+ZQQw3B/uj4xPgKWwEyWygbZJ/hpWwD/yqUftS9vPDEtUKYCfWhnsRUBP4HFgEfAbUCM5tCzwfde2FwOJguyCJ8S3G2p4jv4PPBOceAHyY0+9CkuJ7JfjdmoV90O2fNb7g6x7YqJ0fkxlfsP+lyO9c1Lmp+Pll95mSlN9BL7XhnHMuJm9ics45F5MnCOecczF5gnDOOReTJwjnnHMxeYJwzjkXkycI53IhIrtl7yqzCassKiINoyuJOleYlEl1AM4VAdtUNT3VQTiXbP4E4Vw+BesB/DtYE+A7EflbsL+hiHwRFKP7XEQODPbvK7Y+w8xgOzK4VWkReS6o9/+JiFQIzr8qWAdgloiMStG36UowTxDO5a5CliamvlHHNqjq4cCTwGPBvieAEaraAiuUNzTYPxT4Sq3QYGtsBi5AE+ApVW0OrAdOD/YPAVoF9xkc1jfnXHZ8JrVzuRCRzapaOcb+JUA3Vf0pKKj2m6rWFJHfsfIRO4P9K1S1loisBuqp6h9R92iI1exvEnx9A5CmqneLyEfAZqxi7bsaFCl0Lln8CcK5gtFsXufFH1Gvd5PZN9gTq4vVGpgaVBh1Lmk8QThXMH2j/v02eD0Jqz4KcA4wMXj9OXAZgIiUFpGq2d1UREoB9VX1S+AGoCrwl6cY58Lkf5E4l7sKsvfC9R+pamSoa3URmYU9BZwd7LsSeFFE/gmsBi4I9l8NDBORi7AnhcuwSqKxlAZeDZKIAENVdX3CviPn4uB9EM7lU9AH0VZVf091LM6FwZuYnHPOxeRPEM4552LyJwjnnHMxeYJwzjkXkycI55xzMXmCcM45F5MnCOecczH9P+GmbnV1Xi7XAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_accuracy = history.history['accuracy']\n",
        "dev_accuracy = history.history['val_accuracy']\n",
        "plt.clf() \n",
        "plt.plot(epochs, train_accuracy, 'b', label='train')\n",
        "plt.plot(epochs, dev_accuracy, 'r', label='dev')\n",
        "plt.title('Accuracy curves')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "dYmlmtNhagBy",
        "outputId": "589d9375-1bb1-490c-a4c4-1e3bc8bab7cb"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# appears that we can still improve quite a bit"
      ],
      "metadata": {
        "id": "LbSmigCJagEX"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model = keras.models.Sequential()\n",
        "model.add(keras.layers.Embedding(50000, maxlen_Xtrain)) #limiting to 50 000 words\n",
        "model.add(keras.layers.LSTM(100)) # 100 memory units\n",
        "model.add(keras.layers.Dense(500, activation='relu', name='hidden'))\n",
        "model.add(keras.layers.Dense(num_classes, activation='softmax')) # softmax activation for multi-class classification\n",
        "model.summary()\n",
        "# this model is a bit larger"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UFjwM-7jagIz",
        "outputId": "74216d8d-9fd6-49a8-a145-5b3eaadb6e81"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " embedding_1 (Embedding)     (None, None, 232)         11600000  \n",
            "                                                                 \n",
            " lstm_1 (LSTM)               (None, 100)               133200    \n",
            "                                                                 \n",
            " hidden (Dense)              (None, 500)               50500     \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 25)                12525     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 11,796,225\n",
            "Trainable params: 11,796,225\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "sgd = keras.optimizers.Adam(learning_rate=0.001)\n",
        "model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])"
      ],
      "metadata": {
        "id": "Z50KtgtXeyfV"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "num_epochs = 40 # lets try some more epochs too and see what it does\n",
        "history = model.fit(padded_Xtrain, Y_train, epochs=num_epochs,\n",
        "                     validation_data=(padded_Xval, Y_val), batch_size=100)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IyzePR1Ne302",
        "outputId": "02afba45-ceda-4765-a951-a4936b226509"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/40\n",
            "320/320 [==============================] - 26s 74ms/step - loss: 2.5796 - accuracy: 0.2227 - val_loss: 2.4596 - val_accuracy: 0.2391\n",
            "Epoch 2/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 2.4377 - accuracy: 0.2668 - val_loss: 2.2579 - val_accuracy: 0.3376\n",
            "Epoch 3/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 2.2558 - accuracy: 0.3326 - val_loss: 2.0449 - val_accuracy: 0.4046\n",
            "Epoch 4/40\n",
            "320/320 [==============================] - 23s 70ms/step - loss: 2.0489 - accuracy: 0.3920 - val_loss: 1.7793 - val_accuracy: 0.4764\n",
            "Epoch 5/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.8323 - accuracy: 0.4489 - val_loss: 1.5631 - val_accuracy: 0.5392\n",
            "Epoch 6/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.6457 - accuracy: 0.4960 - val_loss: 1.4193 - val_accuracy: 0.5721\n",
            "Epoch 7/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.4908 - accuracy: 0.5374 - val_loss: 1.2980 - val_accuracy: 0.5955\n",
            "Epoch 8/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.3730 - accuracy: 0.5679 - val_loss: 1.2010 - val_accuracy: 0.6271\n",
            "Epoch 9/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.2877 - accuracy: 0.5912 - val_loss: 1.1334 - val_accuracy: 0.6473\n",
            "Epoch 10/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.2142 - accuracy: 0.6100 - val_loss: 1.0976 - val_accuracy: 0.6575\n",
            "Epoch 11/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.1543 - accuracy: 0.6280 - val_loss: 1.0284 - val_accuracy: 0.6773\n",
            "Epoch 12/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.1007 - accuracy: 0.6466 - val_loss: 0.9806 - val_accuracy: 0.6897\n",
            "Epoch 13/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.0625 - accuracy: 0.6568 - val_loss: 0.9440 - val_accuracy: 0.6981\n",
            "Epoch 14/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 1.0184 - accuracy: 0.6710 - val_loss: 0.9091 - val_accuracy: 0.7108\n",
            "Epoch 15/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.9733 - accuracy: 0.6824 - val_loss: 0.8623 - val_accuracy: 0.7261\n",
            "Epoch 16/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.9319 - accuracy: 0.6973 - val_loss: 0.8403 - val_accuracy: 0.7301\n",
            "Epoch 17/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.9125 - accuracy: 0.7029 - val_loss: 0.8092 - val_accuracy: 0.7402\n",
            "Epoch 18/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.8719 - accuracy: 0.7174 - val_loss: 0.7695 - val_accuracy: 0.7570\n",
            "Epoch 19/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.8406 - accuracy: 0.7250 - val_loss: 0.7517 - val_accuracy: 0.7596\n",
            "Epoch 20/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.8168 - accuracy: 0.7321 - val_loss: 0.7254 - val_accuracy: 0.7705\n",
            "Epoch 21/40\n",
            "320/320 [==============================] - 23s 72ms/step - loss: 0.7900 - accuracy: 0.7425 - val_loss: 0.7002 - val_accuracy: 0.7740\n",
            "Epoch 22/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.7652 - accuracy: 0.7506 - val_loss: 0.6758 - val_accuracy: 0.7826\n",
            "Epoch 23/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.7262 - accuracy: 0.7634 - val_loss: 0.6408 - val_accuracy: 0.7956\n",
            "Epoch 24/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.6987 - accuracy: 0.7736 - val_loss: 0.6198 - val_accuracy: 0.8011\n",
            "Epoch 25/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.6752 - accuracy: 0.7799 - val_loss: 0.5920 - val_accuracy: 0.8101\n",
            "Epoch 26/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.6487 - accuracy: 0.7874 - val_loss: 0.5774 - val_accuracy: 0.8177\n",
            "Epoch 27/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.6404 - accuracy: 0.7916 - val_loss: 0.5734 - val_accuracy: 0.8179\n",
            "Epoch 28/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.6321 - accuracy: 0.7918 - val_loss: 0.5566 - val_accuracy: 0.8233\n",
            "Epoch 29/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.6051 - accuracy: 0.8038 - val_loss: 0.5376 - val_accuracy: 0.8278\n",
            "Epoch 30/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.5850 - accuracy: 0.8078 - val_loss: 0.5135 - val_accuracy: 0.8356\n",
            "Epoch 31/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.5594 - accuracy: 0.8164 - val_loss: 0.4935 - val_accuracy: 0.8392\n",
            "Epoch 32/40\n",
            "320/320 [==============================] - 25s 78ms/step - loss: 0.5385 - accuracy: 0.8233 - val_loss: 0.4780 - val_accuracy: 0.8476\n",
            "Epoch 33/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.5263 - accuracy: 0.8274 - val_loss: 0.4761 - val_accuracy: 0.8458\n",
            "Epoch 34/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.5263 - accuracy: 0.8247 - val_loss: 0.4653 - val_accuracy: 0.8498\n",
            "Epoch 35/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.5323 - accuracy: 0.8244 - val_loss: 0.4633 - val_accuracy: 0.8503\n",
            "Epoch 36/40\n",
            "320/320 [==============================] - 23s 70ms/step - loss: 0.5107 - accuracy: 0.8311 - val_loss: 0.4567 - val_accuracy: 0.8527\n",
            "Epoch 37/40\n",
            "320/320 [==============================] - 23s 70ms/step - loss: 0.4964 - accuracy: 0.8353 - val_loss: 0.4341 - val_accuracy: 0.8581\n",
            "Epoch 38/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.4808 - accuracy: 0.8408 - val_loss: 0.4213 - val_accuracy: 0.8642\n",
            "Epoch 39/40\n",
            "320/320 [==============================] - 23s 71ms/step - loss: 0.4632 - accuracy: 0.8445 - val_loss: 0.4131 - val_accuracy: 0.8647\n",
            "Epoch 40/40\n",
            "320/320 [==============================] - 23s 70ms/step - loss: 0.4509 - accuracy: 0.8492 - val_loss: 0.4024 - val_accuracy: 0.8701\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_loss = history.history['loss']\n",
        "dev_loss = history.history['val_loss']\n",
        "\n",
        "epochs = range(1, len(dev_loss) + 1)\n",
        "\n",
        "plt.plot(epochs, train_loss, 'b', label='train')\n",
        "plt.plot(epochs, dev_loss, 'r', label='dev')\n",
        "plt.title('Loss curves')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Binary Crossentropy')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "2nqYVu0kryg3",
        "outputId": "c6235baa-8321-4c49-f289-8e7bfc794b36"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_accuracy = history.history['accuracy']\n",
        "dev_accuracy = history.history['val_accuracy']\n",
        "plt.clf() \n",
        "plt.plot(epochs, train_accuracy, 'b', label='train')\n",
        "plt.plot(epochs, dev_accuracy, 'r', label='dev')\n",
        "plt.title('Accuracy curves')\n",
        "plt.xlabel('Epochs')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "qKvNthQXryka",
        "outputId": "8b8fe7ff-7588-468a-b90f-aa280eea45f4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deZhT9fX48fdhAFlFVrEsgoIKoiKrWndQURFUVEBRcSnUvVr7q7TWItW69GvdqiIgiIoC7ojUjUWoKAKCgCyCgCyCDpuiMDAw5/fHuSNhyDCZmdwkk5zX8+RJcnOTeybKPbmf5XxEVXHOOZe5yiU7AOecc8nlicA55zKcJwLnnMtwngiccy7DeSJwzrkM54nAOecynCcC55zLcJ4IXMoRkSkisllEDkh2LM5lAk8ELqWISBPgFECBbgk+dvlEHq+0ylq8LnV5InCp5irgM+B54OrIF0SkkYi8ISLZIrJRRP4T8drvRGSRiGwVkYUi0ibYriLSLGK/50XkvuDx6SKyRkT+LCLrgREiUlNExgfH2Bw8bhjx/loiMkJEvgtefyvYvkBELojYr4KIbBCR46P9kSLSXUTmishPIvKNiHQJtq8Ukc4R+w0UkZeCx02Cv+c6EVkFTBKR/4rIzQU++0sRuTh4fJSIfCgim0RkiYhcFrHfecF3tVVE1orInbH9J3LpxhOBSzVXAaOC2zkicjCAiGQB44FvgSZAA2B08NqlwMDgvQdiVxIbYzxefaAWcCjQD/s3MSJ43hjYDvwnYv8XgSrA0UA94NFg+wtAn4j9zgPWqeqcggcUkQ7B/n8CDgJOBVbGGC/AaUAL4BzgFaB3xGe3DGJ/V0SqAh8CLwex9gKeDvYBeA7or6rVgVbApGLE4NKIJwKXMkTkZOwkNlZVZwPfAJcHL3cAfgP8SVV/UdUcVf1f8Nr1wMOqOlPNMlX9NsbD5gF/V9UdqrpdVTeq6uuquk1VtwL3YydeROQQ4Fzg96q6WVVzVfXj4HNeAs4TkQOD51diSSOa64Dhqvqhquap6lpVXRxjvAADg+9gO/Am0FpEDg1euwJ4Q1V3AF2Blao6QlV3BUnpdeDSYN9coKWIHBj8PV8UIwaXRjwRuFRyNfCBqm4Inr/MnuahRsC3qroryvsaYUmjJLJVNSf/iYhUEZFnReRbEfkJmAocFFyRNAI2qermgh+iqt8BnwA9ROQgLGGMKuSYpYkXYHXEcbcC72K/9sGuDvKPeyjQUUS25N+wRFE/eL0HduXyrYh8LCInliImV4Z5Z5NLCSJSGbgMyAra6wEOwE7Cx2Env8YiUj5KMlgNHF7IR2/DmnLy1QfWRDwvWH73j8CRQEdVXS8irYE5gATHqSUiB6nqlijHGoldnZQHPlXVtYXEtL94f4kSb0EFY34F+LuITAUqAZMjjvOxqp4V7UCqOhPoLiIVgJuBsViSchnGrwhcqrgQ2A20BFoHtxbANKzt/3NgHfCgiFQVkUoi8tvgvcOAO0WkrZhmEU0lc4HLRSQr6JA9rYg4qmP9AltEpBbw9/wXVHUd8F+snb1m0CF8asR73wLaALdhfQCFeQ64RkQ6iUg5EWkgIkdFxNsr+Ox2wCVFxAswAfv1PwgYo6p5wfbxwBEicmXweRVEpL2ItBCRiiJyhYjUUNVc4CesmcxlIE8ELlVcDYxQ1VWquj7/hnXUXoH9Ir8AaAaswn7V9wRQ1VextvyXga3YCblW8Lm3Be/LbxZ5q4g4HgMqAxuw0UvvFXj9SqxtfTHwA/CH/BeCNvvXgabAG4UdQFU/B67BOpp/BD7GTuQAf8OuFjYD9wZ/034F/QFvAJ0j9w+ajc7Gmo2+A9YDD2FXWvl/y8qgCez32PfjMpD4wjTOxY+I3AMcoap9itzZuRThfQTOxUnQlHQd9kvbuTLDm4aciwMR+R3WOftfVZ2a7HicK45Qm4aCzrnHgSxgmKo+WOD1Q4HhQF1gE9BHVdfs80HOOedCE1oiCMZdfw2chXXszQR6q+rCiH1eBcar6kgRORO4RlX9sto55xIozD6CDsAyVV0OICKjge7Awoh9WgJ3BI8nU/SIDurUqaNNmjSJb6TOOZfmZs+evUFV60Z7LcxE0ICIGZDYVUHHAvt8CVyMNR9dBFQXkdqquledGBHph9WBoXHjxsyaNSu0oJ1zLh2JSKFlV5LdWXwncJqIzMEm+qzFJhXtRVWHqGo7VW1Xt27UhOacc66EwrwiWMve09UbBtt+FdRnyS+XWw3oUcjUfeeccyEJ84pgJtBcRJqKSEVsduO4yB1EpI6I5McwABtB5JxzLoFCuyJQ1V3BghnvY8NHh6vqVyIyCJilquOA04EHRESxKo83leRYubm5rFmzhpycnKJ3LsMqVapEw4YNqVChQrJDcc6lkTJXYqJdu3ZasLN4xYoVVK9endq1ayMiSYosXKrKxo0b2bp1K02bNk12OM65MkZEZqtqu2ivJbuzOC5ycnLSOgkAiAi1a9dO+6se51zipUUiANI6CeTLhL/ROZd4XnTOOedS2aZNMHMmzJgBF1wAxx8f90OkzRVBMm3ZsoWnn3662O8777zz2LLFR8s65wI7dsDnn8OTT0KfPnDEEVC7NnTpAgMHwmefhXJYvyKIg/xEcOONN+61fdeuXZQvX/hXPGHChLBDc84lkyqsXAn/+x9MmwarVsGuXZCba/eRj3fuhBUr7B7gkEOgY0e49lq7b9sWDjwwlDA9EcTBXXfdxTfffEPr1q2pUKEClSpVombNmixevJivv/6aCy+8kNWrV5OTk8Ntt91Gv379AGjSpAmzZs3i559/5txzz+Xkk09m+vTpNGjQgLfffpvKlSsn+S9zzhVLXh4sWLDnxD9tGqwN5tEedJD9wq9QwW5VqkD58va4fHm7de9uJ/0OHaBhw4SFnXaJ4A9/gLlz4/uZrVvDY48V/vqDDz7IggULmDt3LlOmTOH8889nwYIFvw7zHD58OLVq1WL79u20b9+eHj16ULt27b0+Y+nSpbzyyisMHTqUyy67jNdff50+fXyRK+dS2oYN1pQzY4bdf/YZ5Df3/uY3cMope26tWkG51GyNT7tEkAo6dOiw11j/J554gjfffBOA1atXs3Tp0n0SQdOmTWndujUAbdu2ZeXKlQmL1zlXhNxc+P57+PbbPR23n38Oy5fb6+XKwdFHwyWXwMkn24m/aVMoIyP90i4R7O+Xe6JUrVr118dTpkzho48+4tNPP6VKlSqcfvrpUecCHHDAAb8+zsrKYvv27QmJ1bmMoArr18PXX9sv9tzcfW87d9otO9v2Xbduz/2GDXt/XsOG1oTTv/+e9vtq1UIN/7vvoGpVa2GKt7RLBMlQvXp1tm7dGvW1H3/8kZo1a1KlShUWL17MZyH1+juX9r79Fl5/HebPh5o1oVYtu9WuvedxrVrwyy+wZAksXrz3/U8/xXacihWhfn27HXYYnHSSddzWrw8NGkCbNtbsE5KcHFi0CL78cs9t3jzYuBGefRaCLsa48kQQB7Vr1+a3v/0trVq1onLlyhx88MG/vtalSxcGDx5MixYtOPLIIznhhBOSGKlzZcw339jJ/7XXrEkG7IT88892K0qjRnDkkXDVVXZ/5JGWOPI7bAveKlaE6tVDb9JRtRP70qV2W7bM7ufPt7y1OyjGX7kyHHMMXHQRHHccnHFGOPGkRa2hRYsW0aJFiyRFlFiZ9Le6DKRqv97feMNO/nPm2Pb27a39vUcPOPxw27ZzJ2zebGfUTZvstnEjVKoERx1lI3QimmmTZetWmDXLuhXmz99z8o+cQlSuHDRpAi1b2gn/2GPtvlkzyMqKTxz7qzXkVwTOucTKb/DO/xkceb9sGWzbZvudeCI88ghcfLGdJQuqWBEOPthuKWL3bli40E76M2bYIKKFC21UKUDjxpafeveG5s333Jo2tT8nWTwROOfCt3EjTJgA48bBBx/s3V5foYK1xTdvDmeeab/mzz8/oePoi0vVpgcsXAhffWW3hQvtF39+i1XNmtaP3KPHnqkBBQYLpgxPBM65cHz9tZ3433nHJljl5Vmn62WXWYdrs2Z2a9w4fu0fIVmzBiZNsvlh+Sf9H3/c83qdOjZ6tG9fO+l37Gh/WhkZPeqJwDkXB9u22Rly/nwb5vL++9bWD9bY/de/QrdulgBCmFS1ebN1si5aZKNuGjTYc6tXr/h5ZsMGmDIFJk60BPD117a9Zk1rv7/8cjvx59/K+lLqngicc7FTtXo4c+bYST//tmyZvQY21OXkk+GWW6BrVzj00LgdfssWGzy0aJHd8k/+339f+HuysuxCJD8xVKtmuSgra9/73Fxr18+vTlCtGpx2mk0X6NTJRvCk6OTgUvFE4JwrXHa2zaCdOdPuP//c2vvB2j2aNbOfyFdcYWfJY46x9v44NfXs3AmffgoffQQffmhh5He81qgBLVrAeefZff6tShVrv492W7QItm+3Tt28vH3vwUrK3HefdVe0a2ddGOnOE0FIBg4cSLVq1bjzzjuTHYpze+TkwOTJMH26/fzNy9tzU93z+Pvv7aybX+okv4TChRfaUM62bW2sY5UqcQ1P1drfP/zQbh9/bPPDsrKss/Wvf7Vf6EcfbYOFCmuDP+QQO4m72HgicC7drVsH775rnbYffWTt+SL2U7dcuei3GjXshH/TTXYGbtMmtBIKqlawc+xYGDPGRpKCDbPs2xc6d7aJVDVqhHJ4hyeCuLr//vsZOXIk9erVo1GjRrRt25ZvvvmGm266iezsbKpUqcLQoUM55JBDOPbYY1mxYgXlypXjl19+4aijjmL58uVUyITrUBcuVWvkHjcOxo+32Uxgo3Ouucba7U8/3SZeJdHixXbiHzPGmmzKlbPmmDvvtHVYGjdOangZJdREICJdgMeBLGCYqj5Y4PXGwEjgoGCfu1S1dKu1JKMONTB79mxGjx7N3Llz2bVrF23atKFt27b069ePwYMH07x5c2bMmMGNN97IpEmTaN26NR9//DFnnHEG48eP55xzzvEk4Epn6VJ4+WW7ff21/erv2BHuv99O/scck7TxjDt3WrWIJUtsUNEbb1j9HBE49VTrV+7Rw0b4uMQLLRGISBbwFHAWsAaYKSLjVHVhxG53A2NV9RkRaQlMAJqEFVOYpk2bxkUXXUSVoM20W7du5OTkMH36dC699NJf99uxYwcAPXv2ZMyYMZxxxhmMHj16n9XNnIvJunX2k3rUKPvlL2K/9u+80xY5ScKZ9csvrXshv9bbkiVWrTm/fg5YHbfHH7eqESHWb3MxCvOKoAOwTFWXA4jIaKA7EJkIFMhfe60G8F2pj5oKdagDeXl5HHTQQcyNcoXSrVs3/vKXv7Bp0yZmz57NmWeemYQIXZmTX8Ng+nSrxTNpknXutmkD//d/0LNnUmbkqtqY+wcesJDAWp6aN7dpBD177qn5duSRoa246EoozETQAFgd8XwN0LHAPgOBD0TkFqAq0DnEeEJ16qmn0rdvXwYMGMCuXbt455136N+/P02bNuXVV1/l0ksvRVWZN28exx13HNWqVaN9+/bcdtttdO3alawUn1npkmTduj1Fa2bMsF/9+TUMDj/chtFcfrmVZUiCvDx46y1LALNm2Wid//s/Kw906KHpOeY+HSW7s7g38LyqPiIiJwIvikgrVc2L3ElE+gH9ABqnaA9SmzZt6NmzJ8cddxz16tWjffv2AIwaNYobbriB++67j9zcXHr16sVxxx0HWPPQpZdeypQpU5IYuUs5334LDz5oI31WB7+lype3vqqrr95Tw6B586S1+efmWmvUQw9ZE9Dhh8OQIVbtOWKNJVdGhFaGOjixD1TVc4LnAwBU9YGIfb4Cuqjq6uD5cuAEVf2hsM/1MtSZ87dmnDVr4J//hGHD7ATfvbs1pnfsCMcfH+ooH1W7+Fi40EbwrF1b+BSDXbusftyqVdbsM2CAtfX7RW1qS1YZ6plAcxFpCqwFegGXF9hnFdAJeF5EWgCVgOwQY3Iu9Xz3nbWtDBliZ9zrroO//MUWVQnBTz9ZF0N+8bRFi/Ytolahgl2ERJtiIGITup55Bs49t+wUVnOFCy0RqOouEbkZeB8bGjpcVb8SkUHALFUdB/wRGCoit2Mdx321rK2U41xJrV9vbSuDB1tbyzXXWJt/tNr7paBqo0nffddu06bZ4cAGFbVsaRUiWrSwxy1b7n/Wrks/ofYRBHMCJhTYdk/E44XAb+N0LCTN/8/1HJkGcnOtxMPYsTbef+dOa1i/+26r0RMnO3ZYeYb8k/8339j2li1tqk2XLtask6r18V1iJbuzOC4qVarExo0bqV27dtomA1Vl48aNVErybFBXArt2WU3jsWNtJtXGjVauoVcva2Bv3rzUh9i508buf/yxHeqTT6ySRKVKVp7h9tttrZc4X2y4NJEWiaBhw4asWbOG7Oz07l6oVKkSDVN41SYXYcsWmD0bXn3VFl/fsMFO/t262cIs55xTqs7fHTusEOiUKXbynz7dqmqCTSC+9lo7xJlnxr0unEtDaZEIKlSoQNOmTZMdhss0Gzda43v+WrvLllkbzLJle0o1V60KF1xgJ/8uXaxWfynMnWt9yqNGWaeviFWB7tfPqnKecoqtluVccaRFInAuYVStgueTT1pBt/x+GxGrktasmY2lPPxwm+TVqVOpf5L//DOMHm0JYOZMu5C47DKbtHXqqbZqlnOl4YnAuVhs3QovvAD/+Y/NoKpXz9r3TzrJTv5NmsR9JtUXX+z59f/zzzZk84knoE8fP/m7+PJE4Nz+LF1qJ//nn7e2mPbtLSFcdlkoU2hV4YMP4N57bWWuypWtTk+/fnDCCT6k04XDE4HLbNu325Ta776zW+TjFSts+E2FCnDppXDrrTbLNwT5CWDgQCsr1Lix/fq/8ko46KBQDuncrzwRuMz05Zf2M/vzz/d9rWJFq418yCF2Zu7fH+rXDyWMaAng2WdtZa6KFUM5pHP78ETgMsvOnVbP5/77oVYta4Np3NhO/Pkn/1q1Qm+D8QTgUoknApc5Zs+2Afbz5llNhccfT+jU2rw8G/Xzxhvw5pvW/eAJwKUCTwQu/eXkwKBB8PDDVkRn3Dgb258Aubkwdaqd+N9807oeype3iV4DBlg+8gTgks0TgUtvM2ZYMbdFi+z+3/8OvfdV1Wb7jhxpOWfTJhv9c+65cNFFVurBh3+6VOKJwKUfVRvtkz8Iv0ED+O9/bWZviDZtspGlgwfbOr01alhFiYsvhrPP9lIPLnV5InDpIzvbfoYPG2Zn4urV4ZZbrFkopEVyVe2iY/BgW0M+J8fG+48caSNOS1lRwrmE8ETgyra8PCv5MHQovP22NcqfdBIMH26TvqpWDeWwa9das8+QIVb/p1o16/Dt399WlHSuLPFE4MqOnBwbarN48Z7b//5naybWrg033wzXX29F90M49LRp8P778N57troXWE3/wYNt/fjq1eN+WOcSwhOBS13r1llxt3nz7KS/YoVdAeRr0sTOxP/6l63vG8eSD6rWuvTee3by//hjm4RcsaJV+Lz6autyaNXKyz64ss8TgUs9ublW3+fvf7ez79FHQ7t2Vm3tqKPsdsQRce99jVzVa/x4WL7cth95JPzud1bf/7TTQmttci5pPBG41PLxx3DTTdb2cu65VnCnWbPQDrduHUyYYCf/Dz6AX36xMs+dOsGdd1oIvqqXS3eeCFxqWLcO/vQnG+556KHw1ls29jKkdpe5c+1wH31kzxs1sgJvXbva0o4+1NNlEk8ELrkim4F27IC//Q3uuiu0M/H339s68c89ZyWFBg2y7oVjjvG2fpe5PBG45MjLs7V8Bw6EhQtDbwbKybHSQvffb90Ot99uOcdLPDsH5cL8cBHpIiJLRGSZiNwV5fVHRWRucPtaRLaEGY9LAbt328yrY4+1cf55eVaE5913Q0kCqpZvWra0C40zzrDuh0ce8STgXL7QrghEJAt4CjgLWAPMFJFxqrowfx9VvT1i/1uA48OKxyXZ7t3w6qvwj3/YFUCLFvDKKzb9Nisr7odbt85W+HriCet/btUKPvwQOneO+6GcK/PCbBrqACxT1eUAIjIa6A4sLGT/3sDfQ4zHJcPu3TB2rCWARYvsp/no0bbAe5wSwI4dMGeO1fX/7DNLAKtW2Wt16sAzz9g8s/LeEOpcVGH+02gArI54vgaIus6fiBwKNAUmFfJ6P6AfQOPGjeMbpYu/vDxb+eu11+wqYNUqmwswZowlgHKlb5FUtQuKJ5+0Rd537rTtjRtbrZ8//MHu27QJZWlh59JKqvxG6gW8pqq7o72oqkOAIQDt2rXTRAbmYpSXB9On28n/9ddhzRpb6/fss61B/uKL45IAAGbNgttus8Mdc8yek37HjrbImHOueMJMBGuBRhHPGwbboukF3BRiLC4sc+bYWMw33rCG+QMOsNoLDzxgg/Lj2CO7fj385S/w/PNQt64dtm/fuOUX5zJWmIlgJtBcRJpiCaAXcHnBnUTkKKAm8GmIsbh4mz4d7rvP6vxXqgTnnWfNPl27xr362s6d1uk7aJANA73zTpsLEFJlaecyTmiJQFV3icjNwPtAFjBcVb8SkUHALFUdF+zaCxitqt7kk+pUYdIkSwBTplhP7P33W0mIGjVCOdz48fDHP1rR0a5drZXpiCPifijnMlqofQSqOgGYUGDbPQWeDwwzBhcHqjbO/777bBWW3/wGHn3UKrGFUIFt504bWPTvf8OXX1qNuQQsMOZcxkqVzmKXqqZPhxtvtDNykyZWfL9v31CG4mzcCM8+axUn1q2zgUbDhsFVV1m/s3MuHJ4IXHSqNjbzj3+Ehg1t7cXevUM5I3/9NTz2mHUCb99uA41GjLB7r//jXPg8Ebh9/fyzzcAaM8YqgI4cGUo9hkWLbBTQ229bfunTx2oAtWoV90M55/bDE4Hb2+LFNuZ/yRJ48EGr1Rzn8Zk//GC15oYMsS6Gu++2/uaDD47rYZxzMfJE4PYYOxauuw4qV7bCPGeeGdeP377dmoAeeAC2bYMbboB77rE5Ac655PGpOM7WBLj9dujZ06bqzpkT1ySQl2frzRx1lDUF5VcAffJJTwLOpQJPBJlM1eYDnHGG/VS/9VZ73qBBXD4+L88uLDp2tPb/OnVg8mTrEzjyyLgcwjkXB940lIk2bLAO4CFDbMhOzZpWwa1Xr7h8/A8/2AigoUNh2TIbdPTii3D55V4OwrlU5IkgU6jC1Kk2UP/1123W1m9/C3/9q60JULlyqT4+Lw8mTrTc8vbb1tp06qnWKdyjh1WhcM6lJk8E6S4vzwryP/mkjQSqUQP694d+/eIyTnPjRjv5Dx0KK1ZA7drWwnT99dYn4JxLfZ4I0tnmzdY4P2GC1WkeMcKWh4zTwvATJ9rHr19v3Qz//CdcdJHX/3eurPFEkK7mzbP5AKtWwdNPw+9/H7dpurm5NuzzoYf21AFq3TouH+2cSwJPBOnolVesbaZGDRsFdNJJcfvoFSus0sSMGVZz7rHH4naB4ZxLEh/DkU5yc+GOO2x4Ttu2toZjHJPA6NH2y3/xYpt7NmSIJwHn0oEngnTx/fdw1llWHvrWW60Bv379uHz0L7/YhOPeva0i6Ny5NtDIOZcevGkoHXz+ufUHbNpkA/b79InbR8+dawlgyRIbafr3v3tJaOfSjSeCsm7xYujc2abtTp8et17b3bttNbC777aP/uijuJcecs6lCE8EZdmWLdC9u83WmjIFGjeOy8euWgVXX20fefHF1hdQu3ZcPto5l4I8EZRVu3fDFVfA8uW2jnCcksDo0TbSdPdum3Zw9dW+OIxz6c47i8uqu++2iWJPPgmnnFLqj9uyxboWeveGli1tZcq+fT0JOJcJPBGURWPG2KIx/frZz/dSmjoVjjvOrgYGDbLnhx0Whzidc2VCkYlARC4QEU8YqWLuXLjmGisY9+STpfqoXbtshvDpp0PFitbX/Le/QXlvMHQuo8Rygu8JLBWRh0WkWGXERKSLiCwRkWUiclch+1wmIgtF5CsRebk4n59xsrOtc7h2basgWrFiiT9q9WqrD/SPf1g/wJw50KFDHGN1zpUZRf72U9U+InIg0Bt4XkQUGAG8oqpbC3ufiGQBTwFnAWuAmSIyTlUXRuzTHBgA/FZVN4tIvdL9OWksN9dmcf3wA0ybVqoFft9+2y4qcnPhpZesz9k5l7liavJR1Z+A14DRwCHARcAXInLLft7WAVimqstVdWfw3u4F9vkd8JSqbg6O80Mx488cd9wBH39s9Z7btSvRR+Tk2KTjCy+Epk3tKsCTgHMulj6CbiLyJjAFqAB0UNVzgeOAP+7nrQ2A1RHP1wTbIh0BHCEin4jIZyLSpZAY+onILBGZlZ2dXVTI6efpp+E//7FkUMJZw0uWwIknWrfC7bdbf0CzZnGO0zlXJsXSLdgDeFRVp0ZuVNVtInJdHI7fHDgdaAhMFZFjVHVLgWMNAYYAtGvXTkt5zLJDFR5+GO66C84/3+o+l8CLL8INN9i8s3fega5d4xync65Mi6VpaCDwef4TEaksIk0AVHXift63FmgU8bxhsC3SGmCcquaq6grgaywxuLw8+NOfLAn06gVvvFHs4TyqtlTkVVdZMdK5cz0JOOf2FUsieBXIi3i+O9hWlJlAcxFpKiIVgV7AuAL7vIVdDSAidbCmouUxfHZ6y82Fa6+1Yj833wyjRhV7hNDu3fbWe++1iWETJ9oi8s45V1AsiaB80NkLQPC4yLOSqu4CbgbeBxYBY1X1KxEZJCLdgt3eBzaKyEJgMvAnVd1Y3D8irWzbZgV+Ro602V1PPAHlijeNY8cOW5Lg6afh//0/GD7c5wY45woXy+khW0S6qeo4ABHpDmyI5cNVdQIwocC2eyIeK3BHcHNbtsAFF8Ann9hZ/IYbiv0RW7fausETJ8K//gV33hlCnM65tBJLIvg9MEpE/gMINhLoqlCjykTr1sE551hZ6TFjSrTyS3Y2nHuu9QWMHGl9A845V5RYJpR9A5wgItWC5z+HHlWmWbHCiv1nZ1shuc6di/0RK1fC2WfDmjU2Yez88+MfpnMuPcXUciwi5wNHA5UkKEepqoNCjCtzbNtmM1X99EAAABdnSURBVLx+/BEmT4b27Yv9EfPn28XE9u22gEwclyl2zmWAIhOBiAwGqgBnAMOAS4gYTupKQdX6AebPtyuBEiSBlSuhUydbPnLaNGjVKv5hOufSWyzDUU5S1auAzap6L3AiNszTldaQIfDCC7YQcJeok6r366efbF5Abq6tTeNJwDlXErE0DeUE99tE5DfARqzekCuNmTOt8E+XLlb7uZh277ZFZBYvhvfegyOPDCFG51xGiCURvCMiBwH/Ar4AFBgaalTpbsMGuOQSOOQQK/9ZzHkCYJOOJ0yAZ54pUd+yc879ar+JIFiQZmJQ++d1ERkPVFLVHxMSXTrKX2t4/Xqr/FaCVeGHDoVHH7ULijgsUOacy3D7/SmqqnnYmgL5z3d4Eiile++FDz6waqJt2xb77ZMnw403WovSI4+EEJ9zLuPE0iYxUUR6iPgy5qX27ru2JNg118D11xf77UuXQo8ecMQRtr6wl41wzsVDLImgP1ZkboeI/CQiW0Xkp5DjSj8rVsCVV0Lr1vDUU1DMvLp5s40QysqyUtI1aoQUp3Mu48Qys7h6IgJJazk59lNe1dYarly5WG/PX6VyxQqrIXTYYSHF6ZzLSLFMKDs12vaCC9W4/XjpJVsX8s03S3QWHzDAEsDzz8Mpp8Q/POdcZoullflPEY8rYWsRzwbODCWidDR4MBxzDHQvuGRz0b74wkYI9e8PV18dQmzOuYwXS9PQBZHPRaQR8FhoEaWbWbNg9uwS9Qvs3m3DQ+vWhQcfDCk+51zGK8m4kzVAi3gHkraeeQaqVi3RovPPPmsTkF9+GQ46KITYnHOO2PoInsRmE4ONMmqNzTB2RdmyBV55xZLAgQcW663r11vfQOfOtmSxc86FJZYrglkRj3cBr6jqJyHFk15efNFqQ5dgpbE77rAlJ59+utgtSs45VyyxJILXgBxV3Q0gIlkiUkVVt4UbWhmnas1CHTrA8ccX660ffmgXEgMHQvPm4YTnnHP5YppZDEQOfK8MfBROOGlk2jRYtKjYxYBycqyERPPm8Oc/hxSbc85FiOWKoFLk8pSq+rOIVAkxpvQweLD18PbsWay3PfggLFtmVwWVKoUUm3PORYjliuAXEWmT/0RE2gLbY/lwEekiIktEZJmI3BXl9b4iki0ic4Nb8QvwpKIffoDXXrOB/1Viz5lLlsADD8Dll3tpaedc4sRyRfAH4FUR+Q4QoD5Q5M9cEcnCKpeehQ05nSki41R1YYFdx6jqzcULO8WNGGF1Ifr3j/ktqtYkVLmyVxV1ziVWLBPKZorIUUD+GlhLVDU3hs/uACxT1eUAIjIa6A4UTATpJS/PJgCcdhq0iH26xcsv23KTTz8N9euHGJ9zzhVQZNOQiNwEVFXVBaq6AKgmIjfG8NkNgNURz9cE2wrqISLzROS1YNZytBj6icgsEZmVnZ0dw6GT6MMPrTpcMTqJN2+24aIdOkC/fiHG5pxzUcTSR/C7YIUyAFR1M/C7OB3/HaCJqh4LfAiMjLaTqg5R1Xaq2q5u3bpxOnRInnnGakJcfHHMbxk0yFavHDzYykw751wixZIIsiIXpQna/ivG8L61QOQv/IbBtl+p6kZV3RE8HQYUf8muVLJmjS0WcN11UDGWrwiWL7cyRNdcU+zpBs45FxexJIL3gDEi0klEOgGvAP+N4X0zgeYi0lREKgK9gHGRO4jIIRFPuwGLYgs7RQ0bZr2+xWjfuftuW2ns3ntDjMs55/YjllFDfwb6AfmN3vOwkUP7paq7RORm4H0gCxiuql+JyCBglqqOA24VkW5Y6YpNQN/i/wkpIjfXVpXv0gWaNo3pLbNn2wziAQOgQbTeE+ecS4BYRg3licgM4HDgMqAO8HosH66qE4AJBbbdE/F4ADCgOAGnrPHj4bvvrI8gBqo2c7h2bZ9B7JxLrkITgYgcAfQObhuAMQCqekZiQitjBg+Ghg3hvPNi2v2DD2zVsUcf9fWHnXPJtb8+gsXYKmRdVfVkVX0S2J2YsMqYb7+1M/v111uDfxHy8uwqoEmTEhUmdc65uNrfWetirIN3soi8B4zGZha7gl5+2e6vuiqm3UeNgi+/tPsDDggxLueci4Go6v53EKmKzQjujV0hvAC8qaofhB/evtq1a6ezZs0qesdEUbX1iGvUgE+KXqYhJweOPBLq1LHVx8rFMm7LOedKSURmq2q7aK8VeRpS1V9U9eVg7eKGwBxsJJEDmDcPvvoKrrgipt2fegpWrYKHHvIk4JxLDcU6Fanq5mCWb6ewAipzRo2yfoHLLity182b4f774eyzvbqocy51+G/S0sjLs4kA55xjbT1FePBBW8b4oYcSEJtzzsXIE0FpTJ1qZSViaBZavRoef9zWsW/dOgGxOedcjDwRlMaoUVC1KnTrVuSu99xj/cr/+EcC4nLOuWLwRFBSO3bYKmQXXWTJYD8WLoSRI+GWW+DQQxMUn3POxcgTQUlNmGAN/n36FLnrI4/Y+sN37bNYp3POJZ8ngpIaNQrq1YNO+x9AtX49vPQS9O0bU3+yc84lnCeCkvjxRysy16tXkSUlnnrKCpPefnuCYnPOuWLyRFASr79ufQRFjBbats2KkXbrBs2bJyg255wrJk8EJTFqFDRrBu3b73e3kSNh40b44x8TFJdzzpWAJ4LiWrsWJk+2qwEpvAZfXp6VmG7fHk4+OYHxOedcMcWyQpmLNHq0TQgoolnonXdg6VLbfT/5wjnnks6vCIpr1Cj7mV9Eo/8jj9icgR49EhSXc86VkCeC4li4EObMKfJqYOZMmDYNbrstpnVqnHMuqTwRFMeoUVY7umfP/e72yCNw4IFw3XUJiss550rBE0GsVG0lss6doX79Qnf79lurPNGvnyUD55xLdZ4IYjV9OqxcWWSz0OOPW+fwrbcmJiznnCutUBOBiHQRkSUiskxECq20IyI9RERFJOoyailh1CioXNmKzBXixx9h2DBbo6ZRowTG5pxzpRBaIhCRLOAp4FygJdBbRFpG2a86cBswI6xYSm3HDhgzBrp3h+rVC91t6FDYutUnkDnnypYwrwg6AMtUdbmq7gRGA92j7PcP4CEgJ8RYSmfcONi0Ca65ptBdcnOtWeiMM6BNmwTG5pxzpRRmImgArI54vibY9isRaQM0UtV39/dBItJPRGaJyKzs7Oz4R1qU4cOhYcP9Vhp99VVbrMyvBpxzZU3SOotFpBzwb6DIU6eqDlHVdqrarm7duuEHF2nNGvjgA6sjnZVVSHw2ZPSoo+DccxMbnnPOlVaY053WApFdpg2DbfmqA62AKWI1GOoD40Skm6rOCjGu4nnhBSsc1Ldvobt88gl88QU8+6xNM3DOubIkzNPWTKC5iDQVkYpAL2Bc/ouq+qOq1lHVJqraBPgMSK0koAojRsDpp8Phhxe62+OPQ82aMS1W5pxzKSe0RKCqu4CbgfeBRcBYVf1KRAaJSNGrvaeC//0Pli3bbyfxqlXw5pvwu99BlSoJjM055+Ik1Eo4qjoBmFBg2z2F7Ht6mLGUyPDhNlx0P5XjnnnGLhxuvDGBcTnnXBx5i3Zhtm6FsWNtOcqqVaPusn07DBkCF15olUadc64s8kRQmLFjba3Ja68tdJeXX7bpBV5OwjlXlnkiKMyIETYetGPHqC+rwhNPwLHHwqmnJjg255yLI08E0SxZYmNCr7220OXFpk6FefPsasBXIHPOlWWeCKIZMcImj115ZaG7PPEE1KoFl1+ewLiccy4EnggK2rULRo6E888vdN2Bb7+Ft96yNQcqV05wfM45F2eeCAp6/31Yv36/cweeftqag264IYFxOedcSDwRFDR8ONSrZ1cEUWzbZuWmL7oIGjdOcGzOORcCTwSRsrOt5PSVV0KFClF3GTUKNm/2IaPOufThiSDSqFHWR1BIs1D+kNHWreHkkxMcm3POhSTUEhNliqo1C3XoAEcfHXWXKVNgwQLbzYeMOufShV8R5Js9G+bP3+9M4ieegDp1oHfvBMblnHMh80SQb+RIqFTJagtFsWKFdR/062e7OedcuvBEkO+//4WzzoIaNaK+7ENGnXPpyhMB2Ayxb74pdE3inBx47jkbMtqwYYJjc865kHkiAJg82e7PPDPqy2++aUNG+/dPYEzOOZcgnggAJk6EunWhVauoLz/3HDRpUmiecM65Ms0TgSpMmmRn+ShjQpcvtzxx7bW+ML1zLj35qW3JEvjuu0J/7o8YYfmhb9/EhuWcc4niiWDSJLuP0lG8e7clgnPOgUaNEhyXc84liCeCSZOsetxhh+3z0vvvw9q1cP31SYjLOecSJNREICJdRGSJiCwTkbuivP57EZkvInNF5H8i0jLMePaRl2cjhjp1ito/MGyY9SFfcEFCo3LOuYQKLRGISBbwFHAu0BLoHeVE/7KqHqOqrYGHgX+HFU9UX35pq89H6R/4/nt45x246iqoWDGhUTnnXEKFeUXQAVimqstVdScwGugeuYOq/hTxtCqgIcazr4kT7T5KInjhBStEet11CY3IOecSLszqow2A1RHP1wAdC+4kIjcBdwAVgahDd0SkH9APoHE8V4OZNAmOOgp+85u9Nqva3IGTToIWLeJ3OOecS0VJ7yxW1adU9XDgz8DdhewzRFXbqWq7unXrxufAO3fC1KlRrwamT7dRpX414JzLBGEmgrVA5KDLhsG2wowGLgwxnr3NnAm//BJ12OiwYVCtGlx2WcKicc65pAkzEcwEmotIUxGpCPQCxkXuICLNI56eDywNMZ69TZpkI4VOO22vzT/9BGPHWjXqatUSFo1zziVNaH0EqrpLRG4G3geygOGq+pWIDAJmqeo44GYR6QzkApuBq8OKZx8TJ9qak7Vr77V59GhboN7nDjjnMkWoS1Wq6gRgQoFt90Q8vi3M4xdq2zb49NOoK9A/95ytVNmhQxLics65JEh6Z3FSfPKJdRYX6CiePx8+/9yuBnxNYudcpsjMRDBpEpQvD6ecstfm556DChWgT58kxeWcc0mQmYlg4kTo2HGv3uAdO+DFF20Vsjp1khibc84lWOYlgi1bYPbsfZqF3nrLqk343AHnXKbJvEQwdaoVm4uYP5CXB//6Fxx6KHTunMTYnHMuCUIdNZSSJk6EypXhhBN+3TRypF0kvPiir0LmnMs8mXfamzQJTj4ZDjgAsAlkAwbAiSfCFVckOTbnnEuCzLoi+P57WLBgrzP+ffftKTntQ0adc5kos64IJk+2+6B/YOlSeOwxuOYaaN8+iXE551wSZVYimDQJatSA448H4I47rIXon/9MclzOOZdEmdU0NHGiFZkrX5733oPx4+Ghh6B+/WQH5pxzyZM5VwQrV8Ly5dCpE7m5cPvt0KwZ3JacakfOOZcyMueKYNIkuz/zTJ56ChYvhnHjfh085JxzGStzrgjq1YNLLiG73tEMHAjnnANduyY7KOecS77MuSLo2hW6duXu/rYw2aOP+nBR55yDTLoiAObOhaFD4eabfVF655zLlzGJQNU6hmvXhnvuKXp/55zLFBnTNPTqq1ZvbvBgqFkz2dE451zqyJgrggMPhO7dfS1i55wrKGOuCLp0sZtzzrm9ZcwVgXPOueg8ETjnXIYLNRGISBcRWSIiy0Tkriiv3yEiC0VknohMFJFDw4zHOefcvkJLBCKSBTwFnAu0BHqLSMsCu80B2qnqscBrwMNhxeOccy66MK8IOgDLVHW5qu4ERgPdI3dQ1cmqui14+hnQMMR4nHPORRFmImgArI54vibYVpjrgP9Ge0FE+onILBGZlZ2dHccQnXPOpURnsYj0AdoB/4r2uqoOUdV2qtqubt26iQ3OOefSXJjzCNYCjSKeNwy27UVEOgN/BU5T1R0hxuOccy4KUdVwPlikPPA10AlLADOBy1X1q4h9jsc6ibuo6tIYPzcb+LaQl+sAG0oTd8hSOT6PrWQ8tpLx2EqmNLEdqqpRm1RCSwQAInIe8BiQBQxX1ftFZBAwS1XHichHwDHAuuAtq1S1WymON0tV25U68JCkcnweW8l4bCXjsZVMWLGFWmJCVScAEwpsuyficecwj++cc65oKdFZ7JxzLnnSLREMSXYARUjl+Dy2kvHYSsZjK5lQYgu1j8A551zqS7crAuecc8XkicA55zJc2iSCoiqdJpOIrBSR+SIyV0RmJTmW4SLyg4gsiNhWS0Q+FJGlwX1SFvMsJLaBIrI2+O7mBkOSkxFbIxGZHFTL/UpEbgu2J/27209sSf/uRKSSiHwuIl8Gsd0bbG8qIjOCf69jRKRiCsX2vIisiPjeWic6togYs0RkjoiMD56H872papm/YfMUvgEOAyoCXwItkx1XRHwrgTrJjiOI5VSgDbAgYtvDwF3B47uAh1IotoHAnSnwvR0CtAkeV8cmS7ZMhe9uP7El/bsDBKgWPK4AzABOAMYCvYLtg4EbUii254FLkv3/XBDXHcDLwPjgeSjfW7pcERRZ6dQZVZ0KbCqwuTswMng8ErgwoUEFCoktJajqOlX9Ini8FViEFVFM+ne3n9iSTs3PwdMKwU2BM7GqApC8762w2FKCiDQEzgeGBc+FkL63dEkExa10mmgKfCAis0WkX7KDieJgVc2f3b0eODiZwURxc7B40fBkNVtFEpEmwPHYL8iU+u4KxAYp8N0FzRtzgR+AD7Gr9y2quivYJWn/XgvGpqr539v9wff2qIgckIzYsKoM/w/IC57XJqTvLV0SQao7WVXbYIv03CQipyY7oMKoXXOmzK8i4BngcKA1VorkkWQGIyLVgNeBP6jqT5GvJfu7ixJbSnx3qrpbVVtjhSc7AEclI45oCsYmIq2AAViM7YFawJ8THZeIdAV+UNXZiTheuiSCmCqdJouqrg3ufwDexP4xpJLvReQQgOD+hyTH8ytV/T74x5oHDCWJ352IVMBOtKNU9Y1gc0p8d9FiS6XvLohnCzAZOBE4KChMCSnw7zUiti5BU5uqVUMeQXK+t98C3URkJdbUfSbwOCF9b+mSCGYCzYMe9YpAL2BckmMCQESqikj1/MfA2cCC/b8r4cYBVwePrwbeTmIse8k/yQYuIknfXdA++xywSFX/HfFS0r+7wmJLhe9OROqKyEHB48rAWVgfxmTgkmC3ZH1v0WJbHJHYBWuDT/j3pqoDVLWhqjbBzmeTVPUKwvrekt0rHq8bcB42WuIb4K/JjicirsOwUUxfAl8lOzbgFayZIBdrY7wOa3ucCCwFPgJqpVBsLwLzgXnYSfeQJMV2MtbsMw+YG9zOS4Xvbj+xJf27A47F1iafh51Q7wm2HwZ8DiwDXgUOSKHYJgXf2wLgJYKRRcm6AaezZ9RQKN+bl5hwzrkMly5NQ84550rIE4FzzmU4TwTOOZfhPBE451yG80TgnHMZzhOBcwER2R1RcXKuxLGKrYg0iayq6lwqCXXxeufKmO1q5Qacyyh+ReBcEcTWk3hYbE2Jz0WkWbC9iYhMCoqTTRSRxsH2g0XkzaDO/ZciclLwUVkiMjSoff9BMJsVEbk1WEtgnoiMTtKf6TKYJwLn9qhcoGmoZ8RrP6rqMcB/sKqQAE8CI1X1WGAU8ESw/QngY1U9Dltf4atge3PgKVU9GtgC9Ai23wUcH3zO78P645wrjM8sdi4gIj+rarUo21cCZ6rq8qC423pVrS0iG7CyDbnB9nWqWkdEsoGGakXL8j+jCVbmuHnw/M9ABVW9T0TeA34G3gLe0j018p1LCL8icC42Wsjj4tgR8Xg3e/rozgeewq4eZkZUl3QuITwROBebnhH3nwaPp2OVIQGuAKYFjycCN8CvC5/UKOxDRaQc0EhVJ2N172sA+1yVOBcm/+Xh3B6Vg9Wq8r2nqvlDSGuKyDzsV33vYNstwAgR+ROQDVwTbL8NGCIi12G//G/AqqpGkwW8FCQLAZ5Qq43vXMJ4H4FzRQj6CNqp6oZkx+JcGLxpyDnnMpxfETjnXIbzKwLnnMtwngiccy7DeSJwzrkM54nAOecynCcC55zLcP8fjg9AbSBnFjIAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# better convergence and better scores! accuracy +.10 and loss -.36"
      ],
      "metadata": {
        "id": "y-W-yEt-ryn5"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Y_pred = model.predict(padded_Xtest, batch_size=100)\n",
        "print(Y_pred.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ItkmaSKoB1sx",
        "outputId": "37ec05a1-0062-45e1-c42a-f0ff36b17b58"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(15967, 25)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "y_pred = np.argmax(Y_pred, axis=1)\n",
        "y_pred"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zLnHOKOSH55-",
        "outputId": "b76736e6-e4fd-4c06-924a-91883b31c2af"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([19,  0, 14, ..., 15,  0,  0])"
            ]
          },
          "metadata": {},
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.metrics import accuracy_score, f1_score, classification_report, confusion_matrix\n",
        "print(accuracy_score(y_true=y_test, y_pred=y_pred))\n",
        "print(f1_score(y_true=y_test, y_pred=y_pred, average='macro'))\n",
        "print(classification_report(y_true=y_test, y_pred=y_pred))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "DbAQuhmwrysx",
        "outputId": "3d230b0d-90ee-4ee6-f6e1-feb26cbed7cd"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0.8622158201290161\n",
            "0.8793262635894881\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "         0.0       0.78      0.94      0.85      3597\n",
            "         1.0       0.81      0.85      0.83      2131\n",
            "         2.0       0.85      0.82      0.84      1619\n",
            "         3.0       0.93      0.87      0.90      2241\n",
            "         4.0       0.94      0.78      0.86       251\n",
            "         5.0       0.90      0.84      0.87       341\n",
            "         6.0       0.93      0.86      0.89       232\n",
            "         7.0       0.95      0.80      0.87       449\n",
            "         8.0       0.87      0.83      0.85       527\n",
            "         9.0       0.92      0.81      0.86       489\n",
            "        10.0       0.93      0.79      0.86       397\n",
            "        11.0       0.98      0.72      0.83        71\n",
            "        12.0       0.93      0.79      0.86       282\n",
            "        13.0       0.92      0.86      0.89       257\n",
            "        14.0       0.96      0.85      0.90       184\n",
            "        15.0       0.94      0.84      0.89       403\n",
            "        16.0       0.96      0.89      0.93       139\n",
            "        17.0       0.96      0.86      0.90      1239\n",
            "        18.0       1.00      0.74      0.85        73\n",
            "        19.0       0.84      0.81      0.83       499\n",
            "        20.0       0.96      0.89      0.92       160\n",
            "        21.0       0.99      0.88      0.93        80\n",
            "        22.0       0.96      0.83      0.89       131\n",
            "        23.0       1.00      1.00      1.00         8\n",
            "        24.0       0.94      0.87      0.90       167\n",
            "\n",
            "    accuracy                           0.86     15967\n",
            "   macro avg       0.93      0.84      0.88     15967\n",
            "weighted avg       0.87      0.86      0.86     15967\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        ""
      ],
      "metadata": {
        "id": "rDjwCTvfZlur"
      }
    }
  ]
}