torch_Twitter_Sentiment.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "torch_Twitter_Sentiment.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyMRANR4YL2ePWygiNNvT8p8",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/fractal2k/Twitter-Sentiment-Prediction/blob/master/torch_Twitter_Sentiment.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3tRkFJswJ0DZ",
        "colab_type": "text"
      },
      "source": [
        "# Sentiment140 with PyTorch and GloVe Twitter Embeddings"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8jmPsq3x4HIi",
        "colab_type": "text"
      },
      "source": [
        "[hiddenlayer](https://github.com/waleedka/hiddenlayer) is a convenient visualization library that we use to visualize our training and testing loss/accuracy."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Bq8yHlTKM2jW",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "!pip install hiddenlayer"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5HVZ6s2OJt1u",
        "colab_type": "text"
      },
      "source": [
        "## Kaggle configuration and Getting the dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LYm-AfuPkqGf",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "!mkdir .kaggle"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "O_A_H_Nfg44i",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import json\n",
        "\n",
        "token = {'username': '<your username here>','key':'<your API key here>'}\n",
        "with open('./.kaggle/kaggle.json', 'w') as file:\n",
        "    json.dump(token, file)"
      ],
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2reqUKmfjWt9",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Copy the kaggle directory to root\n",
        "!cp -avr ./.kaggle ~/"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jy7sUb_xjxBe",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "!kaggle config set -n path -v {/content}"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_3W-NAHlg42F",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "!kaggle datasets download -d kazanova/sentiment140 -p /content"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "25aFG_IgmX63",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "!unzip -qq sentiment140.zip"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5nSfCFs_KAlu",
        "colab_type": "text"
      },
      "source": [
        "## Making the necessary imports"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "IFdy7C6GNByL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import re\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "import seaborn as sn\n",
        "from sklearn.model_selection import train_test_split\n",
        "import hiddenlayer as hl\n",
        "\n",
        "from collections import defaultdict\n",
        "\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "import torch.optim as optim\n",
        "from torch.optim.optimizer import Optimizer\n",
        "\n",
        "import torchtext\n",
        "from torchtext import data"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wtmITb1OBYH1",
        "colab_type": "text"
      },
      "source": [
        "## Reading the data and preprocessing"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g2lQ1UPTeEuk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "df = pd.read_csv('training.1600000.processed.noemoticon.csv', names=['target', 'ids', 'date', 'flag', 'user', 'text'], encoding='ISO-8859-1')"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fH7aCk-FoNCM",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "1188c9e1-d2b4-48c9-cfc4-8a5f7bd9a1e1"
      },
      "source": [
        "df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>target</th>\n",
              "      <th>ids</th>\n",
              "      <th>date</th>\n",
              "      <th>flag</th>\n",
              "      <th>user</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>1467810369</td>\n",
              "      <td>Mon Apr 06 22:19:45 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>_TheSpecialOne_</td>\n",
              "      <td>@switchfoot http://twitpic.com/2y1zl - Awww, t...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0</td>\n",
              "      <td>1467810672</td>\n",
              "      <td>Mon Apr 06 22:19:49 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>scotthamilton</td>\n",
              "      <td>is upset that he can't update his Facebook by ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0</td>\n",
              "      <td>1467810917</td>\n",
              "      <td>Mon Apr 06 22:19:53 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>mattycus</td>\n",
              "      <td>@Kenichan I dived many times for the ball. Man...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0</td>\n",
              "      <td>1467811184</td>\n",
              "      <td>Mon Apr 06 22:19:57 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>ElleCTF</td>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0</td>\n",
              "      <td>1467811193</td>\n",
              "      <td>Mon Apr 06 22:19:57 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>Karoli</td>\n",
              "      <td>@nationwideclass no, it's not behaving at all....</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   target  ...                                               text\n",
              "0       0  ...  @switchfoot http://twitpic.com/2y1zl - Awww, t...\n",
              "1       0  ...  is upset that he can't update his Facebook by ...\n",
              "2       0  ...  @Kenichan I dived many times for the ball. Man...\n",
              "3       0  ...    my whole body feels itchy and like its on fire \n",
              "4       0  ...  @nationwideclass no, it's not behaving at all....\n",
              "\n",
              "[5 rows x 6 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "d0hpBnQBoWh1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "raw_data = df[['text', 'target']].copy(deep=True)"
      ],
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RTw2XmuKqkEI",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "outputId": "2b25553a-4598-45cd-e8a2-68a627b828e6"
      },
      "source": [
        "raw_data.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>target</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>@switchfoot http://twitpic.com/2y1zl - Awww, t...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>is upset that he can't update his Facebook by ...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>@Kenichan I dived many times for the ball. Man...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>@nationwideclass no, it's not behaving at all....</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                text  target\n",
              "0  @switchfoot http://twitpic.com/2y1zl - Awww, t...       0\n",
              "1  is upset that he can't update his Facebook by ...       0\n",
              "2  @Kenichan I dived many times for the ball. Man...       0\n",
              "3    my whole body feels itchy and like its on fire        0\n",
              "4  @nationwideclass no, it's not behaving at all....       0"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7V6jdnP6BiUp",
        "colab_type": "text"
      },
      "source": [
        "We sub out the usernames with the ```<user>``` tag, links with the ```<url>``` tag, hashtags with the ```<hashtag>``` tag and numbers with the ```<number>``` tag. After that, we filter out all the punctuations.</br>This is necessary because we use the GloVe Twitter Embeddings, which contain these tags."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nD63Ws7EhBUH",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Sub the rest of the elements before filtering the punctuations out, otherwise \n",
        "# the punctuation regex will get rid of the symbols necessary for these elements\n",
        "# too\n",
        "def clean(text):\n",
        "    USER = '@[\\w_]+'\n",
        "    LINK = 'https?:\\/\\/\\S+'\n",
        "    HASHTAG = '#\\S+'\n",
        "    NUMBER = '\\d+'\n",
        "    PUNCTUATIONS = '[\\.?!,;:\\-\\[\\]\\{\\}\\(\\)\\'\\\"/]'\n",
        "\n",
        "    user_sub = re.sub(USER, ' <user> ', text)\n",
        "    link_sub = re.sub(LINK, ' <url> ', user_sub)\n",
        "    hashtag_sub = re.sub(HASHTAG, ' <hashtag> ', link_sub)\n",
        "    number_sub = re.sub(NUMBER, ' <number> ', hashtag_sub)\n",
        "    clean_text = re.sub(PUNCTUATIONS, ' ', number_sub)\n",
        "\n",
        "    return clean_text.lower()"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UoWOm2w-sMyg",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "raw_data['text'] = raw_data['text'].apply(lambda x: clean(x))"
      ],
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4-TKpnusB4_X",
        "colab_type": "text"
      },
      "source": [
        "The target values are {0, 4}, so we replace 4 with 1 instead."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vkM5MG6nM8wT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "raw_data['target'] = raw_data['target'].replace(4, 1)"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wvjl7B4XB191",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 359
        },
        "outputId": "f2b53b0d-f67a-4fc1-f574-06532ff50298"
      },
      "source": [
        "raw_data.head(10)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>target</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>&lt;user&gt;   &lt;url&gt;    awww  that s a bummer   you...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>is upset that he can t update his facebook by ...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>&lt;user&gt;  i dived many times for the ball  mana...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>&lt;user&gt;  no  it s not behaving at all  i m mad...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>&lt;user&gt;  not the whole crew</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>need a hug</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>&lt;user&gt;  hey  long time no see  yes   rains a ...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>&lt;user&gt;  nope they didn t have it</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>&lt;user&gt;  que me muera</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                text  target\n",
              "0   <user>   <url>    awww  that s a bummer   you...       0\n",
              "1  is upset that he can t update his facebook by ...       0\n",
              "2   <user>  i dived many times for the ball  mana...       0\n",
              "3    my whole body feels itchy and like its on fire        0\n",
              "4   <user>  no  it s not behaving at all  i m mad...       0\n",
              "5                        <user>  not the whole crew        0\n",
              "6                                        need a hug        0\n",
              "7   <user>  hey  long time no see  yes   rains a ...       0\n",
              "8                  <user>  nope they didn t have it        0\n",
              "9                            <user>  que me muera          0"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UEAaAkfWCFuA",
        "colab_type": "text"
      },
      "source": [
        "Splitting into train and test sets:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2OZ0Z9lcwIew",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "train, test = train_test_split(raw_data, test_size=0.15)"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0NXzWBYgDRDB",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "!mkdir data"
      ],
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "euhU5dV4EMv7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "train.to_csv('./data/train.csv', index=False)\n",
        "test.to_csv('./data/test.csv', index=False)"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "il2WNdUBEQob",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "TEXT = data.Field(include_lengths=True)\n",
        "LABEL = data.LabelField(dtype=torch.float)\n",
        "fields = [('text', TEXT), ('label', LABEL)]"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o6EGdFwbFH0r",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "trainds, testds = data.TabularDataset.splits(path='./data', format='csv', train='train.csv', test='test.csv', fields=fields, skip_header=True)"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jUm4n0ArG2SZ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "d9067932-fb65-4833-cfd4-8c9543103d05"
      },
      "source": [
        "TEXT.build_vocab(trainds, max_size=2000000, vectors='glove.twitter.27B.50d', unk_init=torch.normal)"
      ],
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            ".vector_cache/glove.twitter.27B.zip: 1.52GB [12:19, 2.06MB/s]                            \n",
            "100%|█████████▉| 1190835/1193514 [00:41<00:00, 28541.10it/s]"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4_fqah-LSegx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "LABEL.build_vocab(trainds)"
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sjp-Dv21rY-I",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "c412c8d7-a044-4f54-8b03-6521961e5e07"
      },
      "source": [
        "LABEL.vocab.stoi['0'] = 0\n",
        "LABEL.vocab.stoi['1'] = 1\n",
        "LABEL.vocab.stoi"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "defaultdict(<function torchtext.vocab._default_unk_index>, {'0': 0, '1': 1})"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qwU2YlEnSnfn",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "device = torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')\n",
        "BATCH_SIZE = 1024\n",
        "\n",
        "train_iterator, test_iterator = data.BucketIterator.splits(datasets=(trainds, testds), \n",
        "                                                           batch_sizes=(BATCH_SIZE, BATCH_SIZE), \n",
        "                                                           sort_key=lambda x: len(x.text), \n",
        "                                                           sort_within_batch=True, \n",
        "                                                           device=device)"
      ],
      "execution_count": 22,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8xPFFJuoXyfW",
        "colab_type": "text"
      },
      "source": [
        "## Here comes the model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8ISNri5ZN15Y",
        "colab_type": "text"
      },
      "source": [
        "Lookahead optimizer implementation from [Michael Zhang's Github repo](https://github.com/michaelrzhang/lookahead/blob/master/lookahead_pytorch.py)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XO9Qf0cINy-0",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class Lookahead(Optimizer):\n",
        "    r\"\"\"PyTorch implementation of the lookahead wrapper.\n",
        "    Lookahead Optimizer: https://arxiv.org/abs/1907.08610\n",
        "    \"\"\"\n",
        "\n",
        "    def __init__(self, optimizer, la_steps=5, la_alpha=0.8, pullback_momentum=\"none\"):\n",
        "        \"\"\"optimizer: inner optimizer\n",
        "        la_steps (int): number of lookahead steps\n",
        "        la_alpha (float): linear interpolation factor. 1.0 recovers the inner optimizer.\n",
        "        pullback_momentum (str): change to inner optimizer momentum on interpolation update\n",
        "        \"\"\"\n",
        "        self.optimizer = optimizer\n",
        "        self._la_step = 0  # counter for inner optimizer\n",
        "        self.la_alpha = la_alpha\n",
        "        self._total_la_steps = la_steps\n",
        "        pullback_momentum = pullback_momentum.lower()\n",
        "        assert pullback_momentum in [\"reset\", \"pullback\", \"none\"]\n",
        "        self.pullback_momentum = pullback_momentum\n",
        "\n",
        "        self.state = defaultdict(dict)\n",
        "\n",
        "        # Cache the current optimizer parameters\n",
        "        for group in optimizer.param_groups:\n",
        "            for p in group['params']:\n",
        "                param_state = self.state[p]\n",
        "                param_state['cached_params'] = torch.zeros_like(p.data)\n",
        "                param_state['cached_params'].copy_(p.data)\n",
        "                if self.pullback_momentum == \"pullback\":\n",
        "                    param_state['cached_mom'] = torch.zeros_like(p.data)\n",
        "\n",
        "    def __getstate__(self):\n",
        "        return {\n",
        "            'state': self.state,\n",
        "            'optimizer': self.optimizer,\n",
        "            'la_alpha': self.la_alpha,\n",
        "            '_la_step': self._la_step,\n",
        "            '_total_la_steps': self._total_la_steps,\n",
        "            'pullback_momentum': self.pullback_momentum\n",
        "        }\n",
        "\n",
        "    def zero_grad(self):\n",
        "        self.optimizer.zero_grad()\n",
        "\n",
        "    def get_la_step(self):\n",
        "        return self._la_step\n",
        "\n",
        "    def state_dict(self):\n",
        "        return self.optimizer.state_dict()\n",
        "\n",
        "    def load_state_dict(self, state_dict):\n",
        "        self.optimizer.load_state_dict(state_dict)\n",
        "\n",
        "    def _backup_and_load_cache(self):\n",
        "        \"\"\"Useful for performing evaluation on the slow weights (which typically generalize better)\n",
        "        \"\"\"\n",
        "        for group in self.optimizer.param_groups:\n",
        "            for p in group['params']:\n",
        "                param_state = self.state[p]\n",
        "                param_state['backup_params'] = torch.zeros_like(p.data)\n",
        "                param_state['backup_params'].copy_(p.data)\n",
        "                p.data.copy_(param_state['cached_params'])\n",
        "\n",
        "    def _clear_and_load_backup(self):\n",
        "        for group in self.optimizer.param_groups:\n",
        "            for p in group['params']:\n",
        "                param_state = self.state[p]\n",
        "                p.data.copy_(param_state['backup_params'])\n",
        "                del param_state['backup_params']\n",
        "\n",
        "    @property\n",
        "    def param_groups(self):\n",
        "        return self.optimizer.param_groups\n",
        "\n",
        "    def step(self, closure=None):\n",
        "        \"\"\"Performs a single Lookahead optimization step.\n",
        "        Arguments:\n",
        "            closure (callable, optional): A closure that reevaluates the model\n",
        "                and returns the loss.\n",
        "        \"\"\"\n",
        "        loss = self.optimizer.step(closure)\n",
        "        self._la_step += 1\n",
        "\n",
        "        if self._la_step >= self._total_la_steps:\n",
        "            self._la_step = 0\n",
        "            # Lookahead and cache the current optimizer parameters\n",
        "            for group in self.optimizer.param_groups:\n",
        "                for p in group['params']:\n",
        "                    param_state = self.state[p]\n",
        "                    p.data.mul_(self.la_alpha).add_(1.0 - self.la_alpha, param_state['cached_params'])  # crucial line\n",
        "                    param_state['cached_params'].copy_(p.data)\n",
        "                    if self.pullback_momentum == \"pullback\":\n",
        "                        internal_momentum = self.optimizer.state[p][\"momentum_buffer\"]\n",
        "                        self.optimizer.state[p][\"momentum_buffer\"] = internal_momentum.mul_(self.la_alpha).add_(\n",
        "                            1.0 - self.la_alpha, param_state[\"cached_mom\"])\n",
        "                        param_state[\"cached_mom\"] = self.optimizer.state[p][\"momentum_buffer\"]\n",
        "                    elif self.pullback_momentum == \"reset\":\n",
        "                        self.optimizer.state[p][\"momentum_buffer\"] = torch.zeros_like(p.data)\n",
        "\n",
        "        return loss"
      ],
      "execution_count": 24,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uHu2phE56bLo",
        "colab_type": "text"
      },
      "source": [
        "Now, we define the model architecture:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QQJt3i0DTRpk",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class TwitterSentimentPredictor(nn.Module):\n",
        "    def __init__(self, vocab_size, embedding_dim, pretrained_embeddings, hidden_dim, output_dim, n_layers, pad_idx, bidirectional):\n",
        "        super(TwitterSentimentPredictor, self).__init__()\n",
        "\n",
        "        self.embedding = nn.Embedding(vocab_size, embedding_dim, padding_idx=pad_idx)\n",
        "        self.embedding.weight.data.copy_(pretrained_embeddings)  # Initialize embeddings\n",
        "        # We don't freeze the embedding layer because I've found that to give better training accuracy\n",
        "\n",
        "        # Define LSTM Layers\n",
        "        self.lstm = nn.LSTM(input_size=embedding_dim, hidden_size=hidden_dim, num_layers=n_layers, dropout=0.8, bidirectional=bidirectional)\n",
        "\n",
        "        # Final dense\n",
        "        self.dropout = nn.Dropout(0.8)\n",
        "        self.dense1 = nn.Linear(hidden_dim * 2, output_dim)\n",
        "    \n",
        "\n",
        "    def forward(self, input, lengths):\n",
        "\n",
        "        embed = self.embedding(input)\n",
        "\n",
        "        # Pack sequence\n",
        "        packed_embed = nn.utils.rnn.pack_padded_sequence(embed, lengths)\n",
        "        packed_output, (hidden, cell) = self.lstm(packed_embed)\n",
        "\n",
        "        # Take the hidden state of the last layer for output\n",
        "        hidden = self.dropout(torch.cat((hidden[-2, :, :], hidden[-1, :, :]), dim=1))\n",
        "        out = self.dense1(hidden)\n",
        "\n",
        "        return out"
      ],
      "execution_count": 30,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3t_n1ivyU1IS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "INPUT_DIM = len(TEXT.vocab)\n",
        "EMBEDDING_DIM = len(TEXT.vocab.vectors[0])\n",
        "PRETRAINED_EMBEDDINGS = TEXT.vocab.vectors\n",
        "HIDDEN_DIM = 256\n",
        "OUTPUT_DIM = 1\n",
        "N_LAYERS = 2\n",
        "PAD_IDX = TEXT.vocab.stoi[TEXT.pad_token]\n",
        "BIDIRECTIONAL = True\n",
        "\n",
        "model = TwitterSentimentPredictor(INPUT_DIM, EMBEDDING_DIM, PRETRAINED_EMBEDDINGS, HIDDEN_DIM, OUTPUT_DIM, N_LAYERS, PAD_IDX, BIDIRECTIONAL).to(device)"
      ],
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9XkB2cTqVNME",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def calculate_accuracy(output, labels):\n",
        "    preds = torch.round(torch.sigmoid(output))\n",
        "    correct = (preds == labels).float()\n",
        "    acc = correct.sum() / len(correct)\n",
        "    return acc"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AkgtZDdxVO16",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Define loss criterion\n",
        "criterion = nn.BCEWithLogitsLoss()\n",
        "\n",
        "# Define optimizer\n",
        "opt = optim.Adam(model.parameters())\n",
        "opt = Lookahead(optimizer=opt, la_steps=10)"
      ],
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zrAk4yTQVifI",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 314
        },
        "outputId": "bc8669b7-6280-4369-b351-2a1bafecc51a"
      },
      "source": [
        "history = hl.History()\n",
        "canvas = hl.Canvas()\n",
        "EPOCHS = 10\n",
        "iters = 0\n",
        "\n",
        "model.train()\n",
        "# Training loop\n",
        "for epoch in range(EPOCHS):\n",
        "    # Iterate over all the batches\n",
        "    for batch in train_iterator:\n",
        "        # Reset gradients\n",
        "        opt.zero_grad()\n",
        "\n",
        "        # Get the data and lengths from the batch\n",
        "        data, data_lengths = batch.text\n",
        "\n",
        "        # Calculate loss and accuracy\n",
        "        outputs = model(data, data_lengths).squeeze()\n",
        "\n",
        "        loss = criterion(outputs, batch.label)\n",
        "        accuracy = calculate_accuracy(outputs, batch.label)\n",
        "\n",
        "        # Backpropagate the loss\n",
        "        loss.backward()\n",
        "\n",
        "        # Update model parameters\n",
        "        opt.step()\n",
        "\n",
        "        # Update loss plot every 10th batch\n",
        "        if iters % 10 == 0:\n",
        "            history.log(iters, loss=loss, accuracy=accuracy)\n",
        "            canvas.draw_plot([history['loss'], history['accuracy']])\n",
        "        iters +=1\n",
        "\n",
        "print('RIP Computer lol')"
      ],
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "RIP Computer lol\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "siy6WBylb7cb",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 314
        },
        "outputId": "92338517-dc80-4180-bb11-c4d1b597bbde"
      },
      "source": [
        "test_hist = hl.History()\n",
        "test_graph = hl.Canvas()\n",
        "test_iters = 0\n",
        "preds = np.array([])\n",
        "labels = np.array([])\n",
        "acc = []\n",
        "\n",
        "model.eval()\n",
        "with torch.no_grad():\n",
        "    for batch in test_iterator:\n",
        "        data, data_lengths = batch.text\n",
        "\n",
        "        opt._backup_and_load_cache()\n",
        "        outputs = model(data, data_lengths).squeeze()\n",
        "        preds = np.append(preds, torch.round(torch.sigmoid(outputs)).detach().cpu().numpy())\n",
        "        labels = np.append(labels, batch.label.detach().cpu().numpy())\n",
        "        loss = criterion(outputs, batch.label)\n",
        "        accuracy = calculate_accuracy(outputs, batch.label)\n",
        "        opt._clear_and_load_backup()\n",
        "        acc.append(accuracy)\n",
        "\n",
        "        if test_iters % 10 == 0:\n",
        "            test_hist.log(test_iters, loss=loss, accuracy=accuracy)\n",
        "            test_graph.draw_plot([test_hist['loss'], test_hist['accuracy']])\n",
        "        test_iters += 1\n",
        "print(f'Average accuracy: {sum(acc) / len(acc)}')"
      ],
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAA1gAAAEYCAYAAABBWFftAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nOzdd3xUVfr48c9J770ACSn0XiNVEBQsKBYUCyIC1lXcoq4/3bWtuqur665ldRUVEcWKgtgbX0UUkIQeekkgoYX03ub8/jg3yWRIICFlkvC8X6/7mplbz725A/eZc85zlNYaIYQQQgghhBBN5+LsAgghhBBCCCFERyEBlhBCCCGEEEI0EwmwhBBCCCGEEKKZSIAlhBBCCCGEEM1EAiwhhBBCCCGEaCYSYAkhhBBCCCFEM5EASwjRIpRSKUqpSc4uhxBCCCFEa5IASwjR4Sjjn0qpTGv6p1JK1bPuX5RSBXZTsVLKppQKs5YnOyyvUEp9Zre9q1LqCaXUIaVUvlJqg1IqyFrmqZT6j7UsWyn1slLK3W7bAoepUin1orUsTimlHZY/ZLdtiFLqA+v8jiulFiulAuyWP66U2mKV99E6znuGUipVKVWolFqmlAqxW/ajUqrE7rg77ZZdrJRapZTKUUodUUq9rpTyt1t+qut1rlJqvVIqTym1Tyl1ayPKNU8plaiUKlVKLXTY7qTXS5yadU8tta59qlJqxknW/crhWpcppbZYyyKUUu9Z932uUuoXpdRIu20nWN8x++1vtFte7/1nLb9LKbXfuocSlVJn2y17VClV7rDvbnbLpyqltlrzf1VK9bNbdq1SaqdV5mNKqbccvlOn/X0VQpxZJMASQnREtwKXA4OBQcBU4La6VtRa/0Nr7Vc1Af8EftRaH7eW97db5g8cBD6y28XfgDHAaCAAuAEosZbdDyQAA4BewDDgQbtj2x+3E1DssG+AILv1Hreb/wQQDMQD3YFI4FG75XuA+4AvHM9ZKdUfeNUqayRQBLzssNo8u+P2tpsfaB27C9AXiAKesTuneq+XMsHlUuvYgcA1wL+VUoMbWK5D1rEXOJ6TnfquV7uhlHJz0qFfAsow1/564H/W3+QEWuuLHO7fX6m5d/2AdcBwIAR4C/hCKeVnt4tD9ttrrd9yOESd958VqD0FXIW5h94AliqlXO22/cBh3/usbXsCi4HbgSDgM2C53fX+BRirtQ4EugFumPut6pyb8n0VQpxBJMASQrQ4ZWpynrN+0T5kvfe0loUppT63akSylFI/K6VcrGX/TymVrkzN0E6l1HkNPOSNwLNa6zStdTrwLDC7AeVUwCzMA2FdxgNhwMfW+sHAH4FbtNap2tiqta4KsKYCL2its7TWGcALwNx69n0lcAz4uSEniAmslmmt87TWuZjApfphWGv9ltb6KyC/jm2vBz7TWq/UWhcADwHT7Gui6qO1fldr/bXWukhrnQ28BoytZ/Va1wvzsB0AvG1dq3XAdqCqFuGk5dJaf6K1XgZknqqcTaWU+siqoctVSq20DzSUUt5KqWetWp5cZWr0vK1lZ1s1IzlKqYNKqdnW/B+VUjfb7WO2UmqV3WetlLpTKbUb2G3Ne97aR55SKkkpNc5ufVdlal/3Wt+PJKVUV6XUS0qpZx3OZblS6k+nOF9fzD34kNa6QGu9CliOCXZPda3igHHAIgCt9T6t9b+11oe11pVa6/mAB9C7/r00WByQrLVO0lpr65hhQEQDtr0A+FlrvUprXYH5MSUKOMcq98GqH1YslUCPevbV2O+rEOIMIgGWEKI1/BUYBQzB1CqNoKYm5x4gDQjH/HL+F0ArpXoD84CztNb+mIejFKh+iM05yfH6A5vsPm/CLvg4iXGYB7WP61l+I/Cx1rrQ+jwQqACush7Gdyml7nTYRjm8j1ZKBdaz70XWQ6O9VKVUmlLqTWU1W7S8BFyilAq2Ar0rga9OeYZGreujtd6LqbnoZbfOk8o0PfxFKTXhJPsaDyTXs6zW9dJaHwXeA+ZYAcJoIBaoCjQaUq5Tqe96NdZXQE/M/bAeU/NR5V+Y2pkxmKDxPsCmlIq1tnsRcz8PATY24piXAyOpCTjXWfsIAd4FPlJKeVnL7gauA6Zggta5mBq/t4Dr7H6kCAMmAe8qpe5XSn1ez7F7ARVa61128xr6vZmFCVxS6lqolBqCCbD22M2OUEodVaap33+sAM9effffV4CrUmqkVWs1F3ONj9itM9X6sSZZKfU7x+I4vFeYGuaqsp6tlMrF/DBxJfBcPefc2O+rEOJMorWWSSaZZGr2CRMMTbLe7wWm2C27AEix3j8GfAr0cNi+B+YX4kmAeyOPXQn0sfvcE9CAOsV2bwAL61nmA+QBE+zmzbD2+wbgjWmOmAFMtpY/gWl2FI5pUrTWWr+zw75jrTLH283zwzQvdMMEnkuAb+yWdwG+B2zW9B3gUUe53wEedZj3A3C7w7z0qnPDPOT7A56YB8l8oHsd+54MZAO9GnK9rPlTgaOYwLQCU/vXoHLZzXvC8e90quvVxHs5yPq7BWJ+mCwGBtex3gPA0nr28SNws93n2cAqu88aOPcU5ciuOi6wE7isnvW2292D84AvG3CO44AjDvNuwTSXPdW2e4DZ9SwLALYAD9jN64QJIl0wNbErgVftltd7/2ECor8A5db9cxzzI0zVtv2s74YrJgA+DFxnLesDFAITMAHfQ5jvzgN1lDsK0+S2rnu70d9XmWSS6cyapAZLCNEaugCpdp9TrXlg+u/sAb5VJunB/QBa6z2Y5nePAseUUu8rpbrQMAWYB7sqAUCB1trx1+ZqSikfYDr1Nw+cBmQBP9nNK7ZeH9NaF2utNwPvY2oVAP4ObMD8wv4rsAzzYHjUYd83YB6291fN0KaZVqLWukKbmp95wPl2zfg+BHZhHkQDMEHsO/WdnwPH64P1Od869lqtdb7WulSbvjG/2J0TAEqpUZhalat07VqPKidcL6VUH8z1mYV5wO0P3KeUurgh5TqZBlyvBrNq156ymt/lYdWcYpqihQFemOvtqGs98xvqoEM57lVKbbeaIeZgAryqWpGTHestYKb1fibwdgOOfVrXXpkEE50wAYXjMm9MP6c1Wusnq+ZrrY9orbdprW3WPX8fpraoavnJ7r+bgDmYe8fDOr/Pq/5tsPZ7SJumib8Cz2P6a6G13oEJ2P6LCbzCgG2YGvRatGla/DXmfnV0Ot9XIcQZRAIsIURrOIT51bdKjDUP60HqHq11N+BS4O6qvlba9Pc529pWY/pMNEQypililcHU34ytyhWYgODHepbX1SRos/VqP6/6vRV0zdNaR1nnlwkkaa1tDvs+Wb8vx/1W/bs9BPOrf6E2/ZVewSEIOola10eZLGuemICtvmMru/WHYvrnzNVa/1DPNnVdrwHALq31N9bD9U5MEo6LTrNcJ+N4vRpjBnAZpvY0ENPvB8w1OI5JYtK9ju0O1jMfTM2Jj93nTnWsU32trP5W9wFXA8Fa6yAgl5q/w8mO9Q5wmTLJQ/piAvtT2QW4KZMIokpDvjc3Ap9Y92A1ZfpYLsMEL3UmmLGjOfnfyf7+GwJ8rrXeZd1DX2OCpTEN2Bat9RKt9QCtdSjwCOZvu66ebd2o+xqfzvdVCHEGkS++EKI1vAc8qJQKt/olPIxV26KUukQp1cNKMJGLaXpjU0r1VialtyfmgbYY05ynIRZhArUo65fte4CFp9imvj4VKKWigYk4PFRp00foZ+CvyiTy6AtcC3xubRellOqijFGYJkmPOOx7DKY50kcO80da18BFKRWKSZDxozYJLcA8FN6sTMIFb0zmxM1227tb/XVcMA/OXqom09piTD+VcVbfl8cwD8n5SqkgpdQF1vpuSqnrMf2svrb2O8B6f5fWujr9ekOuF6Y2r6f1d1VKqe7AJXblrrdc1n7drHNyxfTD8VJWBrgGXK/G8AdKMQGxD/CPqgVWcLwAk/2wS1VfMus+XQxMUkpdbZU11Op/BKYWc5pSykcp1QNTE3OqMlRgmpy6KaUepnYN0+vA40qpnta1HGSdN1rrNMz98TamD1wxp6BNP7lPgMeUUr5KqbGYILPe2i/rvrsah++WMtkil2C+szc6/qCglJqolIq1yt0VkxXwU2vZSe8/67wuVkp1s7afjOk/ttXa/jJl+iUqpdQI4PdV+7aWD7f+ZuHAfGC5VbOFUup6pVSM9T4WUwNd6weEJnxfhRBnEme3UZRJJpk65kTtPlhemAeOw9b0AuBlLfuTtW4h5tfuh6z5g4DfME2UsjBBSxdr2ThMk7/6jq2Ap63tsqz3ym55ATDO7nMU5mG2Rz37ewDTib+uZVGYh78CYB9wm92y8da5FWH6zFxfx/avYrLqOc6/DthvXZfDmKCxk93yeEzzq0zrHL8GetotX4j5Fd1+mm23fAZwwNr/p0CINT8c8xCbD+QAa7D681jL38QEugV2U3IjrtfVmIfhfOvv/U/A5VTlspY9Wsc5PdqQ69XIe9fPOnY+pjnrLOtYPazl3pjkB+mYHwVWAt529+ZaTP+zg5gAA0xztG+tff5inYtjH6wedp9dMYFcnnU+91H7O+WKSRSz39rnOiDabvuZ1j4n2s37C/DVSc47BFPrVGj9DWbYLTvhO2dd81Qc+jZisvJpzH1vf5+Ms5bfbV27IusavQD4N/D+U5jA+4C1znbgBrvl72G+EwXADuD3DmVbRc2/Ka8CvnbL/o65J6v+LZoPhDbH91UmmWQ6syaldb1dEoQQQgjRDimlxmNqiWO1/EcvhBCtSpoICiGEEB2I1UTvD8DrElwJIUTrkwBLCCGE6CCsfoA5QGfqH8NJCCFEC5ImgkIIIYQQQgjRTKQGSwghhBBCCCGaiZuzDhwWFqbj4uKcdXghhBBCCCGEOG1JSUnHtdbhjvOdFmDFxcWRmJjorMMLIYQQQgghxGlTSqXWNV+aCAohhBBCCCFEM5EASwghhBBCCCGaSYMCLKXUhUqpnUqpPUqp++tYHquU+kEptVkp9aNSKrr5iyqEEEIIIYQQbdsp+2AppVyBl4DJQBqwTim1XGu9zW61fwGLtNZvKaXOBZ4EbmiJAgshhBBCCCEapry8nLS0NEpKSpxdlHbLy8uL6Oho3N3dG7R+Q5JcjAD2aK33ASil3gcuA+wDrH7A3db7/wOWNbjEQgghhBBCiBaRlpaGv78/cXFxKKWcXZx2R2tNZmYmaWlpxMfHN2ibhjQRjAIO2n1Os+bZ2wRMs95fAfgrpUIdd6SUulUplaiUSszIyGhQAYUQQgghhBCnp6SkhNDQUAmuTpNSitDQ0EbVADZXkot7gXOUUhuAc4B0oNJxJa31fK11gtY6ITz8hJTxQgghhBBCiGYmwVXTNPb6NaSJYDrQ1e5ztDWvmtb6EFYNllLKD7hSa53TqJIISFkFbl7QeTC4NqyNpxBCCCGEEKLtaEiAtQ7oqZSKxwRW1wIz7FdQSoUBWVprG/AAsKC5C9rhHd8DCy827919IPosiB1jpqgE8PBxbvmEEEIIIYQ4DX5+fhQUFDi7GK3mlAGW1rpCKTUP+AZwBRZorZOVUo8BiVrr5cAE4EmllAZWAne2YJk7pvULwcUNpr4AhzdB6q/w41OABhd36DLEBFsxYyBmJHgHO7vEQgghhBBCCAcNqcFCa/0l8KXDvIft3i8BljRv0c4gFaWw8V3oPQWGXm8mgOIcOLjWBFsHVsPql+GX5wEFkf0hZnRNLZd/J6eeghBCCCGEECejtea+++7jq6++QinFgw8+yDXXXMPhw4e55ppryMvLo6Kigv/973+MGTOGm266icTERJRSzJ07lz/96U/OPoUGaVCAJVrY9s+gKBMS5tSe7x0EvS4wE0B5MaQlmmAr9RcTlK17zSwLjofYsRA72gReId1AOjQKIYQQQgjL3z5LZtuhvGbdZ78uATwytX+D1v3kk0/YuHEjmzZt4vjx45x11lmMHz+ed999lwsuuIC//vWvVFZWUlRUxMaNG0lPT2fr1q0A5OS0n/QOEmC1BUkLITgO4iecfD13b4gfZyaAynI4stnUcKWuhp1fwsZ3zDK/TlawZdVwRfQDl+ZKGimEEEIIIUTjrFq1iuuuuw5XV1ciIyM555xzWLduHWeddRZz586lvLycyy+/nCFDhtCtWzf27dvHXXfdxcUXX8z555/v7OI3mARYznZ8N6T8DOc90vgAyNUdooabacxdYLPB8Z01TQpTf4XkpWZdr0DoOqqmSWHnIeDm0fznI4QQQggh2qSG1jS1tvHjx7Ny5Uq++OILZs+ezd13382sWbPYtGkT33zzDa+88goffvghCxa0jzx6EmA5W9JCk9xi6Mym78vFBSL6mumsm0BryDlgBVxWLdfub8y6bt4QnVATcEWfBR6+TS+DEEIIIYQQdRg3bhyvvvoqN954I1lZWaxcuZJnnnmG1NRUoqOjueWWWygtLWX9+vVMmTIFDw8PrrzySnr37s3Mmc3wrNxKJMBypvIS04+qz8XgF9H8+1cKgmPNNOQ6M6/gmFW7ZfXjWvkMaJsJ8joPhu7nwuDrILR785dHCCGEEEKcsa644gpWr17N4MGDUUrx9NNP06lTJ9566y2eeeYZ3N3d8fPzY9GiRaSnpzNnzhxsNhsATz75pJNL33BKa+2UAyckJOjExESnHLvN2LIEPr4JblgG3Sc6pwwleXDwNxNspf4Kab+ZgCt2LAy5HvpdBp5+zimbM+QfMTV5nv7OLokQQgghRJNt376dvn37OrsY7V5d11EplaS1TnBcV2qwnCnxTSu5xTnOK4NXAPScZCaAvMOw6T3Y8A58egd8dR/0vxyG3gBdR3bMzIS56bBtmemvlrYOPPwhYTaM/B0ERjm7dEIIIYQQoh2RAMtZMnZB6iqY9Gjbyu4X0BnG3Q1n/8mMwbXhbdi61ARcoT1Mrdbg68x67Vn+Edj2KWz9BA6uMfMiB8LEByFjhxlzbM3/YOB0k0Aksm12ChVCCCGEEG2LBFjOsv4tcHGHIW20w55SEDPKTBf+0wQjG96BH/4GKx6HHpNMYo5eF7WfbIQFx8x5JC8zTSLRJn39xL9C/ysgrGfNuuc9DGtehvWLTI1ej0kw9g8QN65j1uIJIYQQQohmIQGWM5SXwMbFVnKLcGeX5tQ8/WDo9WbK3GvKvvE9+HAW+ITCwKtNsNVpgLNLeqLCTNi+HJI/gZRVpn9ZWC845/+ZoCqiT93bBcfCRf806yW+AWtfhbemmvT2Y38PfS8DV/n6CCGEEEKI2uQJ0Rm2L4fibEiY4+ySNF5od1O7M/GvsPf/TBPCxDdg7f9MFsKhN8CAK8EnxHllLMqCHZ+bPlX7fgJdCSHdYdw9VlDVr+G1UD4hMP7PMPou2Pw+/PoiLJkLQTEwep4JLCW9vRBCCCGEsEgWQWd4cwrkHYK71ret/lenqygLtnxkmhAe2QyunqZ2buhM6DYBXFxbvgzFObDzS9Onat//ga3CJBDpP80EVZ0GNk/TPpvNHOfXF0wfNe9gOOsWGHFr+6iNFEIIIcQZRbIINg/JItiWZew0/X8m/a1jBFdganlG3mamw5tgw2LY8qFplhcQBUNmmCmkW/MetyQPdn5laqr2/gCVZRAYA6PugAHTTHO+5u4v5eICfS8x04G1JtBa+Yx5HXydSYghY4gJIYQQQpyxJMBqbUlVyS2ud3ZJWkbnwWY6/3FT07PhHfj5WROExJ5tarX6XXr6zepKC2DX1yao2v0dVJaaIG7EraamKmp46yWhiBkJMYvh+G5Y/V8zaHTSQlN7N/YP0HVE65RDCCGEEEJQUVGBm5vzw5sOUoXSTpSXwKZ3Te1HR29O5uZpAp6ZH8Mft8K5D0H+IVh2O/yrNyy/ywxw3JAmqmWFJqD64AZ4prsZnDkt0fRhm/ut2f8Ff4foBOdk+AvrCVOfhz9thfH3mmQab0yGNy6AHV+YZoVCCCGEEGewyy+/nOHDh9O/f3/mz58PwNdff82wYcMYPHgw5513HgAFBQXMmTOHgQMHMmjQID7++GMA/Pz8qve1ZMkSZs+eDcDs2bO5/fbbGTlyJPfddx+//fYbo0ePZujQoYwZM4adO3cCUFlZyb333suAAQMYNGgQL774IitWrODyyy+v3u93333HFVdc0eRzdX6IdybZ9qlJbjG8HSa3aIrAKBN4jLsHDqw2tVpbPjYp0EN7mlqtwdeCf6eabcqLTQ1V8lJTY1VeBL4RJolG/ysgZnTba2LpFwHnPghj/2jOcc1L8P4Mc45j7oJB14C7l7NLKYQQQogz1Vf3w5EtzbvPTgPhoqdOudqCBQsICQmhuLiYs846i8suu4xbbrmFlStXEh8fT1ZWFgCPP/44gYGBbNliypmdnX3KfaelpfHrr7/i6upKXl4eP//8M25ubnz//ff85S9/4eOPP2b+/PmkpKSwceNG3NzcyMrKIjg4mDvuuIOMjAzCw8N58803mTt3btOuBxJgta6khaYfUtw4Z5fEOZSC2DFmuuifZjyqjYvh+0fgh8eg52TodQGk/mr6VpUVmDTwg681QVXs2NZJmNFUnn4w6nY462bYtsz0z/rs97DiCdNP7aybTHIMIYQQQrRpx/JKSD6cR5dAb6KCvfHzlEfn0/XCCy+wdOlSAA4ePMj8+fMZP3488fHxAISEmAzU33//Pe+//371dsHBp35mmj59Oq6u5hkxNzeXG2+8kd27d6OUory8vHq/t99+e3UTwqrj3XDDDbzzzjvMmTOH1atXs2jRoiafq9wlreXYDjjwK0x+rO3VvDiDpz8Mu8FMx/fAxndg0/umtso72CSp6D/NBKPtdbwpVzcYeJVJW79/pQm0VjwOP/8bhs2C0XeYdO+iRlkh5B02WSADuoBXgLNLJIQQ4gxUWlHJG6v2898Veygqq6yeH+zjTnSwD9HB3tbkU/3aLgKwBtQ0tYQff/yR77//ntWrV+Pj48OECRMYMmQIO3bsaPA+lF03kJKSklrLfH1r+vY/9NBDTJw4kaVLl5KSksKECRNOut85c+YwdepUvLy8mD59erP04Wrjd0EHsr6DJ7doirAeMOlRmPggHN9l+jS5uju7VM1HKeh2jpmObDVjaa17DX6bbwLJMb+HzoOcXcqWpTWU5pnhCfLSrVfHKR1Kcmpv5+FvAq2ALiaZSUBnu/fWq3ewc/reCSGE6HC01qzYcYzHPt9GamYR5/eLZPaYODILy0jLLiYtu4i07GJ2Hc1nxY5jlFbU7mddFYBFBXnXDsJCzGubD8BaSG5uLsHBwfj4+LBjxw7WrFlDSUkJK1euZP/+/dVNBENCQpg8eTIvvfQSzz33HGCaCAYHBxMZGcn27dvp3bs3S5cuxd/fv95jRUVFAbBw4cLq+ZMnT+bVV19l4sSJ1U0EQ0JC6NKlC126dOGJJ57g+++/b5bzPTP/yq2tvNhkmOs7FXzDnF2atsvVDSL7ObsULavTAJj2Kpz3EKz5n8kqueUjM17YmN9D93PbX7CgtRkLrTpwsl7zD9cOpsoKTtzWN8IESsFxpuloVdDk4lo78Mo7BHtXQMER0A5JQ9y8agdd/p3tAjBrvm+41BwLIYQ4qX0ZBTz2+TZ+3JlB93BfFs0dwfhe9Scl01pzvKCsOuiyD8D2ZBTw465jlJTX/j8ryMfdBF1BPmdUAHbhhRfyyiuv0LdvX3r37s2oUaMIDw9n/vz5TJs2DZvNRkREBN999x0PPvggd955JwMGDMDV1ZVHHnmEadOm8dRTT3HJJZcQHh5OQkICBQV1PFcA9913HzfeeCNPPPEEF198cfX8m2++mV27djFo0CDc3d255ZZbmDdvHgDXX389GRkZzTZemAw03Bo2vQ9Lb4MbP4P48c4ujWhLSnIh8U1Y+4oJSCIHmuaD3kGmFs/FHVw9TPDp6mF9rprn7rCO3TIX9+YJKGyVUJjhUOtU9d4ugKosrb2dcrECnS4ONU7We//OZnLzaFx5Kiug4KhDIOdQC5Z3GGzltbdzcQP/LnZl6HJi2fw6td/mqEIIIU5bQWkFL67YzYJV+/F0c+WPk3py45g43F2b9v+o1vqEmi/HYKwhAVhsmC/jeoThdprlkYGGT23evHkMHTqUm266qd51GjPQsARYrWHBhVBwDO5Kan+1E6J1VJSZmqxfX4SM7c2zT+VqF3C51R2E1fpcFcS5QlFmTS2UraL2fl096q4lsm/G5xvhvGDFZoOi43UEhQ41ahXFtbdTLuAXWVML5h0EXkE1r7XeB9a8b2yQKIQQok2w2TTLNqbz5Fc7yMgvZfrwaP58YW8i/Fsn429jArAR8SG8eN1QIgMaXzYJsE5u+PDh+Pr68t133+Hp6Vnveo0JsOTn2pZ2bLtJTT75cQmuRP3cPGDo9TD4OshLMwGXrRwqy0ytTWUdnyvLTPBTWQaV5WZqyjplBdbyCvAJgbizTwyg/LuYzI5tubmdi4tJme8XAV2G1r2O1mbIhBNq5awascy9pj9Ycc6JgZgjdx8TcFUHYIH1B2OOy9195N8FIYRwgi1puTyyfCvrD+QwODqQ+TcMZ2hM62b4VUoR5udJmJ8nQ7oGnbC8KgBbseMYjy5PZsrzP/PctUMY17ODj6XaypKSkpp9nxJgtbSkheYXf0luIRrCxUUyC7YGpUwQ6RNi+sWdTEWpacpZnFMTdJXk2r13mJ+XDke3mXmleSfft4t77QCsVs2Yl7kflIupjXRxtXtVdcyz5p8wz8V671LHPFfrGI7zXM2/Wz4hJqB2926+ay+EEE6UWVDKv77dyfvrDhLq68HTVw3iqmHRuLi0vR+7qgKwqxO6MiwmmDsWJzFrwW/cNbEHf5jUC9dGlFlrXSsLn2icxrb4kwCrJZUXw6b3rOQWoc4ujRDidLh51tSINZat0iEYq++99bk4G7L3m3mVZWZ7bQNdad7jnCbduHmbQMsn2Lx6W4FXVQDmHVITsFZ99vCV2jkhRJtRUWnjnTWp/Pu7XRSVVTJ3bDx/mNSTAK/2kbW4R4Qfn955Ng9/upUXVuxhXUo2z183pEHNGb28vMjMzCQ0NFSCrNOgtSYzMxMvr4Y3z5QAqyUlLzMPTcPnOLskQghncGbf7HIAACAASURBVHGtCTyag9YnBl3VrzYz1ZpXabeN/Tyb6at2wryq9WxQUQLFWSZDZFGmCf6KMs3nnINmWXF2/WV19bQLuILrCMhCa65N1WdPfwnK2oqqHwc8fM2PDEK0Y7/uPc7flm9j59F8zu4RxiNT+9Ezsu4U322Zt4crz0wfzIj4EB76dCtTnl/FC9cOYUyPk2eojo6OJi0tjYyMjFYqacfj5eVFdHR0g9eXJBct6Y0LTGf7eYny0CCE6Hhslaa2rSjTCsYy7QKyqs9VgVlmTVDmmGq/iot77YArMNqk8A+ONa9BsSYBSVvuA9iWNLZ5a0kOFOee2LzVN7z+bKBV8zx8nHeeQtQjPaeYf3yxnS+2HCY62JsHL+7HBf0jO0Qtzs4j+dyxOIn9xwv5w3m9mHduj0Y1GRTNo0lZBJVSFwLPA67A61rrpxyWxwBvAUHWOvdrrb882T47fIB1bDu8PArOfwLG3OXs0gghRNtgs5kH+KKskwRl1mvOQdOnzb5ppKuH6adYFXBVBWBV771P7CjebmkNZYV2TUhzTnxfb3PThiZoOUlyFq9Ak/zGMSNnXTWXXkF1ZBZ1GBjcM0B+bBStoqS8kld/2sf/ftoDwB0TenDr+G54ubs6uWTNq7C0ggeXbWXphnTO7hHGc9cOIcxPapxb02kHWEopV2AXMBlIA9YB12mtt9mtMx/YoLX+n1KqH/Cl1jruZPvt8AHWl/dB0ptw9w7pfyWEEKerohRy00zftOxUyE6BHOs1O9UEFPa8Ak8MvoLjICgOgro6r7mbzQaluVbwWE9AWd0k065ZpuOYbo48A8E78BQZK4NPXO4VePpDDJQVWQOJ15GFs+p94bETt/Pwq6MGzGGYB5+QjhOEVZSZALU0D0rzHab65hWcON/Ns3Ytrv19HRAt4/fZ0VrzTfIRHv98O+k5xVw8sDMPTOlDdHDHrWHVWvPBuoM8sjyZQG93XrhuKKO6yXNna2lKmvYRwB6t9T5rR+8DlwHb7NbRQID1PhA41LTitnNlRbD5feh7qQRXQgjRFG6eENrdTHUpzqkdcFW9P7Yddn1tkoVUU+ZBvr7aL7/IhjU/rK6FO0ntm32ftarPurLu/bm41e6XFtYDvEeY9ydL+e8ZYPr5tTYPn5P/TcAEF1VB2AmDgR+CfT+a5Y7NRV09a9eA+YSdmAWz6rN9Fsxa8xyyZzYo06bjenbblxU2IDhymF9WYPoxnpIyf0dP/5rJK9A0j636XFZo7uv0JNj2ae2xCZWr1ZTWMQCzJp/QjhOwnsLuo/n87bNtrNpznN6R/rx7y0jGdD9536SOQCnFtSNiGNw1iDsXr2fGa2u45/ze/O6c7m0yM+KZoiE1WFcBF2qtb7Y+3wCM1FrPs1unM/AtEAz4ApO01icklVdK3QrcChATEzM8NTW1uc6jbdn4Liz7Hcz+wowlJIQQovXZbFBwpCb4cqz9ynf4LdDVs3bA5e5l+pBVB09WwFScTb0ZHV3cayf0sO9TdkKSDyv5x5nadK6ywtR01TcYeF66ud6OSVjq68PXmlzcwcs+MLJ77+F34ry61vP0b3y2zcoKc13s72P7+7rQIYmBu++J/Rjtf1joAH3n8krKee673by1OgVfD1funtyLmaNicXM98/pqFpRW8MAnW/hs0yHO6RXOf64ZQojvadZUiwZpShPBhgRYd1v7elYpNRp4Axigdf3/CnboJoJvnG/+U5i37sz8T1MIIdqD8hLIPWg9qKY4BGAHTA1EXdkOT5ai3sNP/t1vaVqfRtbMOoI0+6yZtbJxOmTm9PA9MUBqq5kVywpr1+Q6BmDlRbXX942oPwALiHJODWkD2WyaJUlpPP3NDjILy7j2rBjuPb8XoWd4HyStNYvXHuCxz7YR4uvBf2cMJSGumTLZihM0pYlgOtDV7nO0Nc/eTcCFAFrr1UopLyAMqKMRdgd3dBscXAvn/13+kxVCiLbM3QvCeppJtB9VA23jCq7tYwyjVuPhC5H9zORIayg8bhd8pdQEXwfXwtZPajdjdXE3zQ/DekLvi6xuD22jyd2GA9k8ujyZTWm5DIsJ4s3ZIxgYHejsYrUJSilmjoplSNcg7nx3PdfMX8OfL+jNreO6SZPBVtSQGiw3TJKL8zCB1TpghtY62W6dr4APtNYLlVJ9gR+AKH2SnXfYGqwv/wxJC+Genc039o0QQgghREuqLDcJZRxrvw5vgqy9JqiNHw/9r4C+U53yjHMsv4Snv97JkqQ0Ivw9eWBKHy4fEtUh0q63hLyScu7/eDNfbjnCuX0ieHb6YIKlyWCzamqa9inAc5gU7Au01n9XSj0GJGqtl1uZA18D/DAN0+/TWn97sn12yACrrAie7QO9zocrX3d2aYQQQgghmkZrOLoVkpeaWq7s/SYxS7cJ0H8a9JliMlW2oLIKG2/9msLzP+ymtKKSuWfHc9e5PfHzlAyKp6K1ZtHqVJ74YhsR/l68OGMow2Ja9u91JmlSgNUSOmSAtWExfHoHzP4S4sY6uzRCCCGEEM1Ha1OjlfyJCbhyDpimhD3OMzVbvaeY5B/N6Jc9x3n4063szShkQu9wHr6kH93C/Zr1GGeCzWk53LF4PUdyS7j/oj7cdHa81Pw1AwmwWsPrk8wAj3f+Jv2vhBBCCNFxaQ2H1ptareRlkJdmMnH2mGQFWxeahCBN8O7aA/x12RZiQnx4+JJ+nNsnQoKCJsgtKufPSzbx7bajTO4Xyb+uGkygj/RjbAoJsFraka3wyli44B8w+k5nl0YIIYQQonXYbJCeaGq1kpeZIRDcvKDnZNOMsNcFJgFHI7y2ch9//3I75/QK55WZw/H2aLsZDdsTrTULfknhyS+30ynQi5dmDGNw1yBnF6teWmsOZBWx8WAOlw7u0uYCbAmwWtoX98L6RXDPDkluIYQQQogzk81mshImL4Vty6DgKLj7mCCr/xXQ83xw9653c601//l+Ny/8sJuLB3bmP9cMwcPtzBvTqqWtP5DNXe9u4Fh+CX+d0pcbx8S1ieClrMJG8qFcklKzSUzJJjE1m+MFpQD8fN9Euoa0rbHbJMBqSWVF8Gxvk8Z02nxnl0YIIYQQwvlslXBgtWlGuO1TKDpuBj/ufREMmAbdzzPDJVStbtM8/sU23vwlhenDo3nqykG4SmrxFpNTVMY9H27ihx3HuGhAJ/551SACvFq3yWBOUZkJplKzSUrJZlNaDqUVZhjdmBAfhscGMzw2mIS4YHpF+Le5VPMSYLWkDe/Ap3fCnK8gdoyzSyOEEEII0bZUVkDqKqtmazkUZ4GHv8lC2H8alfETeGD5Tj5MTGP2mDgevqRfm3uY7ohsNs1rP+/j6W92Eh3szUszhjEgqmXGFNNas/94YXUwlXQgmz3HCgBwc1H0jwokITaYBCuoigjwOsUenU8CrJb02nlQmg93rpXkFkIIIYQQJ1NZDvtXmmBr+2dQkkORiy9flA3HdeA0rrjyepSbjNfUmhJTspj37gayCst4aGo/Zo6MaXKTwdKKSram51Y39Vufmk1mYRkAAV5uVs1UCMNjgxkcHdQu+9lJgNVSjmyBV86GC56E0Xc4uzRCCCGEEO1GcXExL7/5BrGHv2Gq53o8KwrMuFp9LjHNCOPGg6uMd9UasgrL+NMHG/lpVwaXDOrMk9MG4t+IJoOZBaUkpWZXT5vTcymzmvvFhfowPDaEhDhTO9Uj3K9D1FDWF2DJHdtUSQtNWtLB1zq7JEIIIYQQ7UZ+STk3LdrIuoPx/P3yl/AcHgl7V9RkI9zwNviEQsxoiOgHEX0hsj+EdJegqwWE+Hrw5uyz+N9Pe3n2250kH8rjpRnD6NflxLHNtNbszSgkKTWLxBQTUO07XgiAu6tiQFQgN46OZXisqaEK9/ds7dNxKqnBaoqyQni2jxlYb9qrzi6NEEIIIUS7kF1Yxo1v/sa2Q3k8e/VgLhsSVXuF8hLY871JjnFoA2TtBW1qQ3D1gLDeENnPCrz6mfcBUR2nq0ZlBeSlQ04q5B0Gr0DwjwS/TuAb3uIB5tp9mdz13gZyi8t59NL+XDE0is1puSSmZlX3n8opKgcg2Med4bHBDIsNJiE2hEHRgXi5t7/mfqdDmgi2hPVvw/J5MOdriB3t7NIIIYQQQrR5x/JKmPnGWlIyi3h5xjAm9Ys89UblJXB8JxzdBseqpu0mCKniGWjVcjkEXt7BLXcyp0trKMqCnBTIToHsVPOaY73mpoGtou5tlQv4hNUEXP6R4N8Z/CLBv1PNPL9IcDv9mqPjBaX88f2NrNpzHFcXRaXNxAzdwnyrM/sNjw2he7hvm0jx7gwSYLWE1841tVh3rOk4v5gIIYQQQrSQg1lFzHxjLRn5pbx+YwJjuoc1bYfF2SbQOrbNCr62w7FkKMmtWce/c+0mhhH9ILz3ScfjahblxSZwqgqaagVRqVCWX3t9nzAIjoXgOAiyXoNjTc1cSR4UHIH8I2Zssfyq90cg/ygUHqup4bPnHWwXcNUVjFmv9QwEXWnTvL06hcN5JQyPMf2nQv3OrOZ+JyN9sJrb4c2QngQXPiXBlRBCCCHEKew5VsDM19dSVFbBOzePZFhMM9QseQebIXLsh8nRGvIO1QRbVbVev62CSjNoLcoFQrrVrumK6A8h8eDSwOZttkpznDoDqBQTCNlz864JmuLOrh1EBcWCp9/pXwdbJRQerwm4ar1aU+YvpkyVZSdu7+FfOwjz6wT+nXD178TszpEwqI+ZLxpEAqzTlbQQ3LwkuYUQQgghxClsTc9l1oLfcFGKD24bTd/OJyZOaDZKQWCUmXpOqplfWQHZ++Fock0zw6PJJlU8VosuNy9Tu+XYxDDnwIkBVM5BsJXbHdcFAqJNwNRzMgTF1QRQwXGm71RL/Sjv4mrVTkVC55Osp7Wp9bOv/XJ8PbTBLC8vqr1teB+IPwe6nQOxY8E7qGXOpQOQJoKno7TAJLfoewlc8YqzSyOEEEII0WYlpmQxZ+E6/D3deOfmkXQLb0JNTUsoK6q7f1f+4RPX9Q6pHTTZ10IFdgXXhqc1b9O0NmO8Fhw11+HQBtj3ExxYbQIv5QJdhtYEXF1HtnyTyzZI+mA1p/WLYPldMPcbiBnl7NIIIYQQQrRJP+/O4NZFSXQK9OKdm0cSFdSOHsKLskywVZILQTEmmPJqwZq39qCiFNISYf9PJuBKTzTJOFw9oesIE2zFTzDB1xmQSl8CrOY0f6LpuHjHaul/JYQQQghRh6+3HuH3722gW7gvb9808owbC+mMUJoPqatrAq6jW8x8zwDTjLDbOaaWK6Jvh3xmliQXzeXwJji0Hi56ukPeKEIIIYQQTbV0Qxr3frSZgVGBvDVnBIE+HaTpnKjN0x96nW8mMIk29q+sCbh2fWXm+0ZA/PiagCs41nllbgUSYDVWVXKLQVc7uyRCCCGEEG3O26tTeOjTZMZ0D2X+rAT8POVx84zhGwYDppkJTHKQfT/VBF1bl5j5wXEm0Iofb179wp1W5JYgd3xjlBbA5o+g/7S2OWidEEIIIYQTvfzjHp7+eieT+kbw3xnD8HJvYMpz0TEFxcCwG8ykNWTsrKndSl4G698y60X0r6ndihtrasbaMQmwGmPrx2ZQuOGznV0SIYQQQog2Q2vNM9/s5OUf93Lp4C48e/Vg3F1dnF0s0ZYoBRF9zDTyNpM2//Am2P+jCbgSF8Cal0G5QtTwmoCr6whwa1/99yTJRWPMn2Cyp/zuV+l/JYQQQogWdSCziKyiMgZFBeLi0nafO2w2zaOfJbNodSrXjYjhicsH4NqGyyvaqPISOLi2pobr0HrQNjNAc8wouOy/EBjt7FLWIkkumurQRjMGwEXPSHAlhBBCiBaVll3E5S//QlZhGaG+HkzoHcF5fSMY1zMMf6+2kzCiotLGfUs288mGdG4d340HLuqDkuckcTrcvUytVbdz4DxMevyUX0z/rQOrwSfM2SVsMAmwGippoYmgJbmFEEIIIVpQcVklt72dRHmFjScuH8C6lCy+336Uj9en4eaiGBEfwrl9Iji3T4RTB+0trajk9+9t4Jvko9wzuRfzzu0hwZVoPl6B0GeKmdoZCbAaojQftnxkMqJ4Bzm7NEIIIYTooLTW3P/JZrYdzuONGxM4t08kM0fFUlFpY/2BHFbsOMaKHUd54ovtPPHFduLDfKuDrbPiQvBwa51+T0VlFdz2dhI/7z7OI1P7MWdsfKscV4j2QAKshtj6MZQVSHILIYQQQrSo137ex6cbD/HnC3pzbp/I6vluri6MiA9hRHwI91/Uh4NZRfzfzmP8sP0Yb69J5Y1V+/HzdGN8rzAm9o5gYp8IwvxaJjFAbnE5Ny1cx/oD2Tx91SCuTujaIscRor2SAKshEt806SOjz3J2SYQQQgjRQa3clcFTX+1gysBO3DGh+0nX7Rriw6zRccwaHUdRWQW/7MlkxY6jrNhxjC+3HEEpGBwdVF271b9LQLM038ssKGXWgt/YdTSf/84YxpSBnZu8TyE6GgmwTuXQBji8Eab8S5JbCCGEEKJFpGYWctd7G+gZ4c8zVw1uVDDk4+HG5H6RTO4Xidaa5EN5rNhxjB92HOM/3+/i39/tolOAFxOtYGtsj1B8PBr/CHg4t5iZr68lPaeY12YlMKF3RKP3IcSZoEHfLqXUhcDzgCvwutb6KYfl/wEmWh99gAitdcforFSV3GLgdGeXRAghhBAdUGFpBbcuSgJg/qzh+Hqe/u/fSikGRAUyICqQ35/Xk4z8Un7ceYwVO47x2aZDvPfbATzcXBjTPZRz+0QwsXcEXUN8Trnf1MxCrn99LTlF5SyaO5IR8SGnXUYhOrpTfoOVUq7AS8BkIA1Yp5RarrXeVrWO1vpPduvfBQxtgbK2vtJ82LIEBlwpyS2EEEII0ey01tz70SZ2H8vnrbkjiA31bdb9h/t7Mj2hK9MTulJWYWNdShY/bDeJMh7+NBlIpnekPxP7mDTwQ7sG4eYwQPCuo/nMfH0t5ZU23r1lJIOi5ZlIiJNpyE8kI4A9Wut9AEqp94HLgG31rH8d8EjzFM/JtiyR5BZCCCGEaDEv/7iXr7Ye4a9T+jKuZ3iLHsvDzYWxPcIY2yOMh6f2Y19GgZWV8Biv/7yPV37aS5CPO+f0CufcPhGc0yucA1lFzFrwGx6uLnxw22h6Rfq3aBmF6AgaEmBFAQftPqcBI+taUSkVC8QDK+pZfitwK0BMTEyjCuoUSW9C5ACIPmGAZiGEEEKIJlmx4yj/+nYnlw3pws3jWj/NebdwP7qF+3HzuG7klZTz867j/LDjKD/uzODTjYdwUeDu6kK4vyfv3jyKmNBTNyUUQjR/kotrgSVa68q6Fmqt5wPzARISEnQzH7t5pa+Hw5skuYUQQgghmt3ejAL+8N5G+nUO4Klpg5w+QG+AlzsXD+rMxYM6U2nTbErLYcX2YxzOLeHPF/SmU6CXU8snRHvSkAArHbAf4CDamleXa4E7m1qoNiFpIbj7wKCrnV0SIYQQQnQgeSXl3LIoEXc3F169YTjeHq7OLlItri6KYTHBDIsJdnZRhGiXGjLc9zqgp1IqXinlgQmiljuupJTqAwQDq5u3iE5Qkmclt5gGXoHOLo0QQgghOgibTXP3BxtJzSzipRnDiA6WZndCdDSnDLC01hXAPOAbYDvwodY6WSn1mFLqUrtVrwXe11q37aZ/DbF1CZQXwvC5zi6JEEIIITqQ537Yzffbj/HwJf0Y3T3U2cURQrSABvXB0lp/CXzpMO9hh8+PNl+xnEhrSHwTIgdC1DBnl0YIIYQQHcTXW4/wwg+7mT48mlmjY51dHCFEC2lIE8Ezy6H1cGQzJMyW5BZCCCGEaBa7juZzz4cbGdw1iMcvH+D0pBZCiJYjAZajquQWA6c7uyRCCCGE6AByi8q5dVEi3h5uvDpzOF7ubSuphRCieUmAZa8kD7Z8DAOulOQWQgghhJNorVmXksW3yUew2dp31+5Km+au9zeQnlPMKzOHSbpzIc4AzT0OVvu25UOT3CJhjrNLIoQQQpxx8krKWbo+ncVrU9l1tACAs3uE8cz0QXQO9HZy6U7PM9/sZOWuDP5xxUAS4kKcXRwhRCuQAKuK1pC4EDoNhC6S3EIIIYRoLVvSclm8NpVPNx6iuLySQdGB/PPKgZRVav7xxXbO/89KHr9sAJcN6dKu+i59tukQr/y0lxkjY5gxMsbZxRFCtBIJsKqkr4ejW+Dif0tyCyGEEKKFFZdV8tmmQyxem8qmtFy83F24bHAU14+KYVB0UPV643qEcc9Hm/jjBxv5bttRnrh8AMG+Hk4secNsO5THn5dsIiE2mEen9nd2cYQQrUgCrCpHNoFXkCS3EEIIIVrQ7qP5LF57gI/Xp5FfUkHPCD8endqPK4ZFE+jtfsL6cWG+fHjbaF75aS/Pfb+L31KyePqqQUzsHeGE0jdMVmEZt76dSJC3By/PHIaHm3R5F+JMopw1LnBCQoJOTEx0yrHrVVYEHjKiuhBCCNGcSisq+Sb5KO+sSeW3/Vm4uyouGtCZmaNiOSsuuMHN/pIP5XL3B5vYeTSfGSNj+OuUvvh6tq3fiisqbcxa8BuJqdl8dNtoBncNOvVGQoh2SSmVpLVOcJzftv5VcjYJroQQQohmczCriMVrD/BR4kEyC8uICfHh/ov6cNXwaML8PBu9v/5dAvl03lj+/d0uXvt5H7/sOc6/rx7M8Ni2kzziH1/u4Ne9mTxz1SAJroQ4Q0mAJYQQQrRBZRW2dtm0rKLSxoodx1i89gArd2eggEl9I7l+VCzjeoTh4tK0fs5e7q78ZUpfzu0TwT0fbmL6K6u5/Zzu/HFSL6dfr4+T0ljwy35mj4ljekJXp5ZFCOE80kRQCCGEcLKswjK2pOeyJS2HzWm5bE3P5VBuCVFB3vTp5E+fzv706RRA387+xIX64uba9gKvo3klvP/bQd5fd4DDuSVEBnhy7VkxXDuia4ulWM8vKefxz7fxYWIafTsH8Nw1Q+jdyb9FjnUqm9NyuOqV1QyPCWbRTSNwb4N/IyFE86qviaAEWEIIIUQryi0qZ0t6LpvTc9iSlsvmtFzSc4qrl8eH+TIwKpC4UB9SMovYcSSPvRmFVFoD7nq4udAzwq864OrTKYA+nf1Pq8ldU9lsml/2HmfxmgN8t/0olTbNuJ5hzBwVy3l9IlotEPw2+QgPfLKF/JIK7r2gFzed3Q3XJtaUNUZGfimX/ncVLkqxfN5YQp3wtxBCtD4JsIQQQohWlldSztb0XBNIWa8Hsoqql8eG+jAgKpBBUYEMjA5kQFQgAV4nZtIrrahkz7ECdhzOZ+fRfLYfzmPHkXwy8kur1wnz8zS1XZ386dM5gD6d/OkR4YeXu2uzn1dWYRlLkg7y7toDpGQWEeLrwfSEaGaMiCE21LfZj9cQxwtK+csnW/h221FGxIfw7PTBdA1p+b7VZRU2rn99DVvSc1ly+xgGRAW2+DGFEG2DBFhCCCFECyoorSA5PdfUTqWZ1/3HC6uXRwd7M8gKogZFBTEgKoAgn6aN53S8oJSdR/LZcSSfHVbQtetoPqUVNgBcXRTxYb706eRPXyvo6tM5gC6BXo0esFdrTVJqNovXHuCLLYcpq7AxIi6E60fFcOGATni6NX8g11haa5YkpfG3z7ahteaRqf2ZnhDdooMTP7RsK2+vSeX5a4dw2ZCoFjuOEKLtkQBLCCGEaCZFZRUkH8pjS1pVQJXDvuOFVP2X2iXQi4HRgQyKDmJAVCADowIJaaXBcSsqbaRkFlmBVx7bD5vXtOyaZoj+Xm5WbVdAdf+u3p388asj5Xl+STnLNqSzeO0BdhzJx9/TjWnDorh+VCy9Ip3T3+lU0rKLuPejTazZl8WkvpE8OW0g4f7N32zv/d8OcP8nW7htfDcemNK32fcvhGjbJMASQgghTkNxWSXbDuexJS2HLel5bEnPYc+xAqwuUUQGeDIwKohB0aaZ38CoQKf0hzqV/JJydh3Nrw64dhzOZ+eRfPJLK6rX6Rribfp2dfKne4Qfa/Zl8enGdIrKKhkQFcDMkbFcOqQLPh5tPwmxzaZZ8Mt+nv5mJ/6ebvxj2kAu6N+p2faflJrNtfNXM6pbKAvnjGjVPl9CiLZBAiwhhBCiAcorbXyTfISVuzLYnJbL7mMF1Qkmwvw8TSAVFVj9GhHg5eQSnz6tNek5xeywgq7tR0zQtS/DBJBe7i5cOrgLM0fFMii6fY7ptOtoPn/6YCPJh/K4clg0j1zar85+bo1xNK+ES15chbe7K8vnjW1yU08hRPskAZYQQghxErnF5bz/2wHe+jWFQ7klhPh6VAdRJqAKIjLAs0X787QVJeWV7MsoJCrYm0DvpgUjbUFZhY0XftjNyz/uoXOgN/+aPpjR3UNPa1+lFZVc8+oadh3NZ+kdY52WFl4I4XwSYAkhhBB1OJBZxIJf9vNh4kGKyioZ1S2Em8/uxrl9Ipo8KK5oW9YfyObuDzaSklnETWfH8+cLejcqy6LWmv/38WY+TEzjlZnDuHBA5xYsrRCirasvwGr7jaiFEEKIZqa1JjE1m9d/3se3247iqhSXDu7C3LPjJc12BzYsJpgv/zCOf3y5nTdW7Wflrgz+c82QBv/N31mTyoeJadx1bg8JroQQ9ZIaLCGEEGeM8kobX245zIJV+9mUlkuQjzvXj4xh1ug4IttxXyrReD/tyuDPH20iq7CMP5zXk99N6H7SgZHX7svk+tfXck6vcF6blSC1m0IIaSIohBDizFXVv2rhrykczi2hW5gvc86O58phUe0iI55oGTlFZTy4bCufbz7M0Jgg/n31EOLDThwoOT2nmEtfXEWgjzvL7hzb5CQZQoiOQQIsIYQQZ5zUzELe/CWlun/V6G6h3Dwunom9pX+VqLF80yEeXLqF8krNX6b0Yeao2OpkJiXllVz1yq+kHi9i6Z1j6RHh5+TSyUPiUgAAIABJREFUCiHaCumDJYQQ4oygtWZdiulf9d32o7i5KKYO7sJNZ8fTv4v0rxInunRwF0bEhfDnJZt46NNkvtt+jKevHERkgCcPfLKF5EN5vD4rQYIrIUSDSA2WEEKIDqGqf9Ubq/az2epfNXNkLDeMjpX+VaJBtNa8syaVv3+5HU83V87rG8En69O5Z3Iv7jqvp7OLJ4RoY6QGSwghRIeUW1TOe+vM+FWHc0voFu7LE5cP4Mph0Xh7NDwFtxBKKW4YHcfYHmH86cNNfLI+nQv7d+LOiT2cXTQhRDsiAZYQQggAisoqSM0sIjWzkLziCqKDveka4kPnQK+TZldzlpTjhbz5y34+SkqjqKySMd1D+fsVA5jQS/pXiabpFu7Hx7eP5qddGYzpHib3kxCiURoUYCmlLgSeB1yB17XWT9WxztXAo4AGNmmtZzRjOYUQQjSD3KJyUrMKScksIvV4IalZJqBKySwiI7+0zm3cXBRdgryJCfGha4gPMQ5ToE/rZVTTWvPb/ixeX7Wf763+VZcOjuKms+Pp1yWg1cohOj43VxfO6xvp7GIIIdqhUwZYSilX4CVgMpAGrFNKLddab7NbpyfwADBWa52tlIpoqQILIYSon9aa4wVlHMgqJOW4CZ5Ss4pMQJVZSE5Rea31IwM8iQ31ZWLvcGJDfYkN9SEu1Bd/LzfSs4s5kFVUPR3MLuab5CNkFZbV2keAlxsxoT4nBGBdg33oEuSNh1vTa7/KK218sdn0r9qSnkuwjzvzJvbghlGxREj/KiGEEG1IQ2qwRgB7tNb7AJRS7wOXAdvs1rkFeElrnQ2gtT7W3AUVQghh2GyaI3klpGQWciCzJniqat5XWFZZva6Lgqhgb2JDfLl4YGfiQn2JsYKomBCfk/ZRig31ZUwd8/NLyjmYZYKvg3YB2I7D+Xy/7RhllbZax+8c6F1T4xVaOwgL9nGvToddl9yict79zfSvOpJXQvdwX/5+xQCmDZX+VUIIIdqmhgRYUcBBu89pwEiHdXoBKKV+wTQjfFRr/bXjjpRStwK3AsTExJxOeYUQ4oxQUWkjPaeYlMwiDlhN+KqDqKwi/n97dx4fVXn2f/xzZScrW1iSQNiXAIZ9UepWq2AVUCtCweWnLWqlrT62LrVPfap2+T2ttlWpllqrNSKguEBdUNxFAgQEJIRNICxhCYQlkH3mfv7IiJGCBJjkTJLv+/XilZkzZ+Z8B25P5vK+zzUVVV8VMZHhRoeWsaS3jGVo55Z0ahV7dDYqrUVsUGaQakqIiSQjJfK4S/L8fsfu4jK27iv5jwLsvXV7/mMZYnx0RKDg+qoIS2sZS/Nmkbzy2Q5ezNlOaaWPkd1a87sr+3Fej2RdDyMiIiEtWE0uIoDuwPlAGvCRmfVzzh2ouZNzbjowHarbtAfp2CIijUbOliJ++0Yeq7YfpMr/1WkyJjKMTq3i6JIcxwW92hxdytexZfUyvPAQKTrCwoz2Sc1on9SMYV1a/cfjJRVVbN9ferQA+7II21R4hA/WFVJeo3CMCg9jTP/q76/q3V7XV4mISMNQmwJrB9Chxv20wLaatgOLnXOVwGYzW091wbU0KClFRBq5nQdL+f2ba3ltRQHtk2KYcm4XOrWOo1OrODq1iiU5Ifobl9I1FLFREfRom0CPtgn/8ZhzjsLD5WwrKmHnwTKGdm5JmwRdXyUiIg1LbQqspUB3M+tMdWE1ATi2Q+CrwETgn2bWmuolg5uCGVREpDEqq/Tx1MebmPb+F/ic4ycXduOW87sSG9X0vkXDzGiTEKOiSkREGrST/gZ3zlWZ2VRgPtXXVz3tnMs1sweAHOfc3MBjF5vZGsAH/Nw5t68ug4uINGTOOd5es5uHXl/DtqJSRvVpx33f7U2HlrFeRxMREZEzYM55cynU4MGDXU5OjifHFhHx0obdxTzw7zV8vGEvPdrGc//lfTinW2uvY4mIiMgpMLNlzrnBx25vemtQREQ8crC0kj8vWM+/FuUTFxXO/1yeweTh6USEB7fLn4iIiHhHBZaISB3z+R2zc7bxh/nr2F9SwcShHbnzOz1oFR/tdTQREREJMhVYIiJ1KGdLEf8zL5fVOw4xpFML7r98KH1Tk7yOJSIiInVEBZaISB3YdbCM372Zx2srCmiXGMOjEwdw+VntG0WrdRERETkxFVgiIkFUVunjH59sZtr7G6nyO358YTdubaJt10VERJoi/cYXEQkC5xzvrNnNQ6/nsbWohEv6tOW+SzPo2Ept10VERJoSFVgiImdo455ifj2vuu169zbxZN00jJHd1XZdRESkKVKBJSJymg6WVvKXBRv416ItNIsK5/5A2/VItV0XERFpslRg1XCkvIq4aP2VSNN1sLSSjzcU0iI2ioz2ibSIi/I6Ukjy+R0vBtquF5VUMGFIR352sdqui4iIiAqso57Lzud/31rLwnsuJDEm0us4IvWmpKKKBXl7mLeygA/XFVLh8x99LCUphoyUJDJSEumTkkhG+0TSWjRr0p3warZdH5zegmfHqO26iIiIfEUFVkBmWhLFZVW8snwH15/dyes4InWqosrPR+sLmbuygAV5uymp8NE2MZprR6Rzab92lFT4WFNwiNyCQ6zZeYj31u7G76qfmxgTESi4kshon0if1ES6Jsc3+mVxuw6W8fs383g10Hb9LxP6MyYzpUkXmyIiIvKfVGAFnJXWnLPSksjKzue6Een60CSNjs/vyN60j7krCnhz9U4OlVXRIjaScQNSGZOZwpBOLQkP+2rcf6t78tHbpRU+1u76quDKLThEVnY+5VXVs11REWH0bJtwtODKaJ9I7/aJjWLJ7bFt16deUN12vTG8NxEREQk+fUKoYfKwdO6as4olm4sY1qWV13FEzphzjs+2HWDuigJe/3wnhcXlxEWFc3GfdozJTGFk99a1mnlqFhXOgI4tGNCxxdFtVT4/m/ceOVpwrSk4xNtrdjErZxsAZtCpVRwZgaWFfVISyUhJpE1CTJ2932ByzrEgbw8P/nsNW4tKuDijLb/8rtqui4iIyDcz55wnBx48eLDLycnx5NgnUlrhY9hvF3BezzY8NnGA13FETotzjrW7ipm7soB5KwvYvr+UqIgwLuzZhjH9U7iwVxtiIsPr7Ni7DpUdXV6YW3CQNTsPsa2o9Og+yQnRXyu4+qQkkd4ylrCw0Jk1rtl2vVubeO6/PONrM3oiIiIiZrbMOTf42O2awaqhWVQ4Vw1KIys7n8LiDJIT1BFMGo4te48wb2UBc1cWsGHPYcLDjJHdWnPHRT24uE9bEuqheYuZ0T6pGe2TmvHt3m2Pbj9YWknezkNfu65r4UebqApc2BUXFU7v9olHm2m0TYyhyueo8vup8Dkqq/xU+vxU+mvc9lU/VuX78r6jwuensspPlf+r218+Vun75tsVged9uT0hJoJfXZbBtSPUdl1ERERqTzNYx9i45zAXPfIhP7+kJ7dd0M3rOCLfaNfBMv69qrqoWrX9IABDO7fk8swULu3bLqTbhpdX+diw+zBrjl7XdZC8ncUcLq86pdeJDDciw8Nq/LFjfn51Oyqi+n5EmBEZEUZU4LGI8K9uR4aHkRATyfjBaSH99yciIiLe0gxWLXVrE8+ILq14YclWbjmv69cu+hcJBUVHKnjj853MXVnA0i1FOAf9UpO479LeXJbZnvZJzbyOWCvREeH0TU36Wotzv9+xtaiEfUfKiQoPJ+LLwig8jMgIIyLsq9tfFkpqSCMiIiKhRAXWcUwens5tM5bz0fpCLujVxus4IhSXVfJ27m7mrSrgkw17qfI7urWJ546LenDZWe3pkhzvdcSgCAszOrWOo1PrOK+jiIiIiJwWFVjH8Z2MtrSOjyYrO18FlnimrNLH+2v3MHdlAe+t3UN5lZ/U5s34wbe6MCYzhd7tEzR7IyIiIhJiVGAdR1REGBOGdGDaBxvZvr+EtBZqyyz1w+93fLihkHkrC3g7dzeHy6toHR/NxKEduTwzhYEdm6uoEhEREQlhKrBOYOKwjvz1g428sGQrP7+kl9dxpIn49bxcnl2UT0JMBJf2a8eYzFSGd2lJhLrYiYiIiDQIKrBOILV5My7s1YZZS7fx02/3ICpCH3Clbs3P3cWzi/K5bkQ69323N9ERdfNdVSIiIiJSd1Q1fINJw9PZe7iC+bm7vI4ijVzBgVLuemkV/VKT+OV3M1RciYiIiDRQKrC+wXndk+nQshlZ2fleR5FGzOd33D5rBVU+P49OHKDZUhEREZEGTJ/kvkFYmPH9oeks3lzEht3FXseRRurx9zayZHMRD47rS2e1JxcRERFp0FRgncT4wWlEhYfx/OKtXkeRRmjpliL+8u56rhiQypUD07yOIyIiIiJnSAXWSbSKj2Z0v3bMWbadkooqr+NII3KgpIKfvvAZHVrG8uC4vl7HEREREZEgqFWBZWajzGydmW00s3uO8/gNZlZoZisCf34Q/KjemTw8neLyKuauKPA6ijQSzjnumfM5e4rLeWziAOKj1dBTREREpDE4aYFlZuHANGA0kAFMNLOM4+w6yznXP/DnqSDn9NTg9Bb0bJtA1uJ8nHNex5FGYMaSrbyVu4u7RvXkrLTmXscRERERkSCpzQzWUGCjc26Tc64CmAmMrdtYocXMmDy8I6t3HGLl9oNex5EGbv3uYh6Yt4ZvdW/ND0Z28TqOiIiIiARRbQqsVGBbjfvbA9uOdZWZrTKzl8ysQ1DShZBxA1KJjQpXy3Y5I2WVPqbOWE5CTAQPj88kLMy8jiQiIiIiQRSsJhfzgE7OubOAd4Bnj7eTmU0xsxwzyyksLAzSoetHQkwk4wakMm9lAQdKKryOIw3UQ6+vYf3uwzw8vj9tEmK8jiMiIiIiQVabAmsHUHNGKi2w7Sjn3D7nXHng7lPAoOO9kHNuunNusHNucHJy8unk9dTkYemUV/l5adl2r6NIA/TW6l1kZW9lyrldOK9Hwxv/IiIiInJytSmwlgLdzayzmUUBE4C5NXcws/Y17o4B8oIXMXRkpCQyoGNzZizeqmYXckp2HCjl7jmrOCstiZ9d3NPrOCIiIiJSR05aYDnnqoCpwHyqC6fZzrlcM3vAzMYEdvuJmeWa2UrgJ8ANdRXYa5OHpbNp7xEWfbHP6yjSQFT5/NwxcwVVPj+PThhAVIS+fk5ERESksarVJz3n3BvOuR7Oua7Oud8Etv3KOTc3cPte51wf51ymc+4C59zaugztpe+e1Z7msZFkLVazC6mdx9/fyJItRTx0RV86tY7zOo6IiIiI1CH9r/RTFBMZztWD0ng7dzd7DpV5HUdC3JLNRTz67gauHJDKFQPSvI4jIiIiInVMBdZp+P6wdKr8jplLt518Z6k1v9+xfOt+fP7GcX3bgZIKbp/5GR1bxvLAuL5exxERERGReqAC6zR0bh3Ht7q35oUlW6ny+b2O0yhU+vzcMXsFV/71U374rxyKyyq9jnRGnHPcPWcVhYfLeWziQOKjI7yOJCIiIiL1QAXWaZo0LJ2dB8t4b+0er6M0eGWVPm7NWs5rKwq4tF87PlxfyBV//ZQte494He20Pb94K/Nzd3PXJb3ol5bkdRwRERERqScqsE7TRb3b0C4xhqzFW72O0qAdKa/ixmeWsiBvNw+M7cNfJw3iuRuHsvdwOWOnLeTTjXu9jnjK1u0q5sF/r+G8HsncNLKz13FEREREpB6pwDpNEeFhTBjagY/WF5K/r+HOtHjpQEkFk55azOLNRTwyPpPrRnQC4OxurXnttnNokxDNtU8v4V+LtjSY7x0rrfDx4xeWkxATyR+vziQszLyOJCIiIiL1SAXWGZgwpCPhYcYMzWKdsj3FZUyYns2agkNM+/5Arhz49Q576a3iePlHZ3NBz2R+9Vou9726moqq0L/e7aHX17B+92EeGZ9JckK013FEREREpJ6pwDoD7ZJi+E7vtszO2UZZpc/rOA3G9v0ljH9yEfn7Snj6hiGM6tvuuPslxETyt2sHc+v5XZmxeCuT/7GYfYfL6zlt7b35+U6eX7yVm8/twrk9kr2OIyIiIiIeUIF1hiYPT2d/SSVvrt7pdZQG4YvCw4x/chFFRyrI+sEwRnZv/Y37h4cZd4/qxZ+v6c+KbQcYO20heTsP1VPa2ttxoJS756wiMy2JOy/u6XUcEREREfGICqwzdHbXVnRuHUdWtpYJnkxuwUHGP7mICp+fmVNGMCi9Ra2fO25AKi/ePIKKKj9XPfEp83N31WHSU1Pl83P7zM/wO3h04gCiIvSflYiIiEhTpU+CZygszJg0rCPL8veH5MxKqFiWX8SE6dlER4Qx++YRZKQknvJrZHZozrwfj6R7m3hufm4Zj727ISSaXzz23kaWbtnPQ+P6kt4qzus4IiIiIuIhFVhBcNXANKIiwsjKzvc6Skj6ZMNeJj+1hNbx0bx469l0SY4/7ddqmxjDrJtHMLZ/Cg+/s54fv/AZpRXeXf+WvWkfj723gasGpjFuQKpnOUREREQkNKjACoIWcVFcdlZ7Xv1sB4fLq7yOE1Lm5+7ixmeWkt4qltk3jyC1ebMzfs2YyHD+fE1/7h7Vi9c/38n4vy1i58HSIKQ9NfuPVHDHrBWkt4rj12P71PvxRURERCT0qMAKksnD0zlS4ePVz3Z4HSVkvLx8Oz96fjl9UhOZNWVEUNuWmxm3nt+Vv187mE2Fh7n8sYUs37o/aK9/Ms457p6zir2Hy3l0wgDioyPq7dgiIiIiErpUYAXJgA7NyWifSFZ2fkhcF+S15xZt4b9mr2RY55Zk3TSMpNjIOjnORRlteeW2c4iNCmfC37KZs2x7nRznWFnZ+by9Zjd3j+pFv7SkejmmiIiIiIQ+FVhBYmZMHp7O2l3F9TqTEoqmvb+R/34tl4t6t+XpG4YQV8ezOz3aJvDabecwKL0Fd764kt++kYfPX3dF7tpdh3jw9TzO75nMjed0rrPjiIiIiEjDowIriMb2TyE+OqLJtmx3zvH7N9fyh/nrGNc/hScmDyQmMrxejt0iLop/3TSU60akM/2jTdz4zFIOlVUG/TilFT5+POMzEmMi+ePVmYSFWdCPISIiIiINlwqsIIqLjuDKgam8vmonRUcqvI5Tr/x+xy9fXc2TH37BpGEdeWR8fyLD63d4RYaH8cDYvvzmir4s3LiXcdMWsnnvkaAe48HX17Bhz2H+dE0mreODd02ZiIiIiDQOKrCCbPLwdCp8fl7M2eZ1lHpT6fNzx+wVPL94K7ec15WHxvX1dGZn0rB0sn4wjP1HKhj7+Cd8vKEwKK/75uc7mRF4j9/qnhyU1xQRERGRxkUFVpD1aJvA0E4tmbFkK/46vA4oVJRV+rg1azmvrSjgrlE9uWd0L8y8XzY3vEsr5k4dSfukZtzwz6X8c+HmM2o+sn1/CXfPWUVmh+bceXGPICYVERERkcZEBVYdmDS8I/n7Svh4416vo9SpI+VV3PjMUhbk7ebBsX340fndvI70NR1axjLnR2dzYa82/HreGu6Z8zkVVf5Tfp0qn5/bZ67A7+CxCQPqfemjiIiIiDQc+qRYB0b1bUeruCiysvO9jlJnDpRUMOmpxSzeXMSfrsnk2hGdvI50XPHREfxt8iCmXtCNWTnbmPRUNnsPl5/Sazz67gZy8vfzmyv60rFVbB0lFREREZHGQAVWHYiOCGf8kA68m7ebggOlXscJuj3FZUyYns2agkP8ddJArhiQ5nWkbxQWZvzskp48OnEAq7YfZOzjC1lTcKhWz130xT4ee38j3xuUxtj+qXWcVEREREQaOhVYdeT7QzvigJlLGlfL9u37Sxj/5CLy95Xw9A1DuKRPO68j1dqYzBReuuVsfH7HVU98ylurd37j/vuPVHDHrBV0bhXHr8f0qaeUIiIiItKQqcCqIx1axnJ+j2RmLt1Gpe/Ur/sJRV8UHmb8k4soOlJB1g+GMbJ7a68jnbJ+aUnMnXoOPdslcEvWcv6yYMNxm5E457hrziqKjlTw6MQBdf5lySIiIiLSOKjAqkOTh6ezp7icd9bs9jrKGcstOMj4JxdR4fMzc8oIBqW38DrSaWuTGMPMKcO5cmAqf1qwnqkvLKekoupr+zyXnc87a3Zz9+he9E1N8iipiIiIiDQ0KrDq0Pk925DavFmDb3axLL+ICdOziY4IY/bNI8hISfQ60hmLiQzn4asz+cWlvXhz9S6+98QidgSul8vbeYiHXs/jgp7J3HhOJ2+DioiIiEiDUqsCy8xGmdk6M9toZvd8w35XmZkzs8HBi9hwhYcZE4d24NMv9vFF4WGv45yWTzbsZfJTS2gdH82Lt55Nl+R4ryMFjZkx5dyuPH39ELYVlTD28U/4ZMNefvzCZyQ1i+QPV2eGxHd6iYiIiEjDcdICy8zCgWnAaCADmGhmGcfZLwH4KbA42CEbsvFDOhARZjyf3fCaXczP3cWNzywlvVUss28eQWrzZl5HqhMX9GrDK7edTXx0BJP/sZgvCg/z52v60zo+2utoIiIiItLA1GYGayiw0Tm3yTlXAcwExh5nvweB/w+UBTFfg9cmIYZL+rbjpWXbKK3weR2n1l5evp0fPb+cPqmJzJoyguSExl1sdGuTwKu3ncNlZ7XnF6N7c063htfAQ0RERES8V5sCKxXYVuP+9sC2o8xsINDBOfd6ELM1GpOHpXOorIp/ryrwOkqtPLdoC/81eyXDOrck66ZhJMVGeh2pXjSPjeLx7w/kh+d28TqKiIiIiDRQZ9zkwszCgEeAO2ux7xQzyzGznMLCwjM9dIMxvEtLurWJJ2tx6C8TnPb+Rv77tVwu6t2Wp28YovbkIiIiIiKnoDYF1g6gQ437aYFtX0oA+gIfmNkWYDgw93iNLpxz051zg51zg5OTk08/dQNjZkwa1pGV2w6wesdBr+Mc18HSSu59eRV/mL+Ocf1TeGLyQGIiw72OJSIiIiLSoNSmwFoKdDezzmYWBUwA5n75oHPuoHOutXOuk3OuE5ANjHHO5dRJ4gbqyoFpNIsMD7mW7c45Xv1sB99++ENmLd3Gzed14ZHx/YkMVwd/EREREZFTddJP0c65KmAqMB/IA2Y753LN7AEzG1PXARuLpGaRjMlM4bUVBRwqq/Q6DgBfFB5m0lOLuX3WClJbNGPu1JHcO7o3YWFqTS4iIiIicjpqdYGNc+4N4I1jtv3qBPuef+axGqfJw9OZlbONl5dt54ZzOnuWo6zSx7T3N/K3DzcRExnGQ+P6MnFoR8JVWImIiIiInBF1MKhH/dKSyExLImvxVq4/u5MnX2L7wbo9/Oq1XLYWlXDFgFR+cWnvRt+CXURERESkvuhCm3o2aXg6G/ccZvHmono97q6DZfzo+WXc8M+lRIYbM344jD9d01/FlYiIiIhIEKnAqmeXn5VCYkxEvTW7qPL5eerjTXz74Q94N28PP7+kJ2/+9FzO7qov0hURERERCTYtEaxnzaLC+d6gDjyXvYXC4vI6nUFavnU/972ymrydh7igZzK/HtOXjq1i6+x4IiIiIiJNnWawPDBpeEcqfY7ZOdvq5PUPlFRw78ufc9UTn3KgpIInJw/k6RuGqLgSEREREaljmsHyQNfkeEZ0acWMxVu55byuQeve55zj5eU7+O0beRworeSmczpz+3d6EB+tf2YRERERkfqgGSyPTB6ezo4DpXywbk9QXm/jnmImTM/mzhdXkt4qlnlTR/LLyzJUXImIiIiI1CN9+vbIxX3akpwQTVZ2Pt/u3fa0X6e0wsdj723g7x9vIjYqgt9d2Y9rBnfQlwWLiIiIiHhABZZHIsPDmDCkA4+/v5FtRSV0aHnq10e9m7eb++fmsn1/Kd8blMa9o3vRKl5t10VEREREvKIlgh6aOLQjBsxYsvWUnldwoJSbn8vhpmdzaBYZzqwpw/nj1ZkqrkREREREPKYZLA+lNG/Ghb3aMnvpNu64qAdREd9c71b6/Pxz4Wb+vGADfue4e1QvbhrZ+aTPExERERGR+qECy2OTh3dkQd5u3srdxZjMlBPul7OliF++upq1u4q5qHcb7r+8z2ktKxQRERERkbqjAstj53ZPpmPLWLKy849bYO0/UsHv31zLrJxtpCTFMP3aQVzcp50HSUVERERE5GRUYHksLMz4/rCO/P7NtazfXUyPtgkA+P2Ol5Zv53dv5FFcVsXN53Xhp9/uTmyU/slEREREREKVLt4JAVcPSiMqPIzns/MBWLermGumL+Kul1bRrU08r//kW9w7ureKKxERERGREKdP7CGgVXw0l/Zrx8vLdxAVEcY/F24hISaC/73qLL43KE3faSUiIiIi0kCowAoRk4en8+qKAv7+8WauGdyBe0b3okVclNexRERERETkFKjAChGD0lvwwNg+ZLRPZHCnll7HERERERGR06ACK0SYGdeN6OR1DBEREREROQNqciEiIiIiIhIkKrBERERERESCRAWWiIiIiIhIkKjAEhERERERCRIVWCIiIiIiIkGiAktERERERCRIVGCJiIiIiIgEiQosERERERGRIDHnnDcHNisE8j05+Im1BvZ6HUKaPI1DCQUahxIKNA4lFGgcyomkO+eSj93oWYEViswsxzk32Osc0rRpHEoo0DiUUKBxKKFA41BOlZYIioiIiIiIBIkKLBERERERkSBRgfV1070OIILGoYQGjUMJBRqHEgo0DuWU6BosERERERGRINEMloiIiIiISJCowBIREREREQkSFVgBZjbKzNaZ2UYzu8frPNJ0mNkWM/vczFaYWU5gW0sze8fMNgR+tvA6pzQuZva0me0xs9U1th133Fm1RwPnx1VmNtC75NKYnGAc/o+Z7QicE1eY2aU1Hrs3MA7Xmdkl3qSWxsTMOpjZ+2a2xsxyzeynge06H8ppU4EFmFk4MA0YDWQAE80sw9tU0sRc4JzrX+N7Nu4B3nXOdQfeDdwXCaZngFHHbDvRuBsNdA/8mQJlgz2sAAAEsUlEQVQ8UU8ZpfF7hv8chwB/CpwT+zvn3gAI/F6eAPQJPOevgd/fImeiCrjTOZcBDAduC4w1nQ/ltKnAqjYU2Oic2+ScqwBmAmM9ziRN21jg2cDtZ4FxHmaRRsg59xFQdMzmE427scC/XLVsoLmZta+fpNKYnWAcnshYYKZzrtw5txnYSPXvb5HT5pzb6ZxbHrhdDOQBqeh8KGdABVa1VGBbjfvbA9tE6oMD3jazZWY2JbCtrXNuZ+D2LqCtN9GkiTnRuNM5Uurb1MDyq6drLJHWOJQ6ZWadgAHAYnQ+lDOgAkvEeyOdcwOpXnZwm5mdW/NBV/1dCvo+BalXGnfioSeArkB/YCfwsLdxpCkws3hgDnC7c+5Qzcd0PpRTpQKr2g6gQ437aYFtInXOObcj8HMP8ArVS152f7nkIPBzj3cJpQk50bjTOVLqjXNut3PO55zzA3/nq2WAGodSJ8wskuri6nnn3MuBzTofymlTgVVtKdDdzDqbWRTVF9HO9TiTNAFmFmdmCV/eBi4GVlM9/q4P7HY98Jo3CaWJOdG4mwtcF+ieNRw4WGPpjEhQHXM9yxVUnxOhehxOMLNoM+tMdZOBJfWdTxoXMzPgH0Cec+6RGg/pfCinLcLrAKHAOVdlZlOB+UA48LRzLtfjWNI0tAVeqT6/EwHMcM69ZWZLgdlmdhOQD4z3MKM0Qmb2AnA+0NrMtgP3A7/n+OPuDeBSqpsKlAD/r94DS6N0gnF4vpn1p3pJ1hbgZgDnXK6ZzQbWUN357TbnnM+L3NKonANcC3xuZisC236BzodyBqx6WamIiIiIiIicKS0RFBERERERCRIVWCIiIiIiIkGiAktERERERCRIVGCJiIiIiIgEiQosERERERGRIFGBJSIiDYKZ3WdmuWa2ysxWmNkwM7vdzGK9ziYiIvIltWkXEZGQZ2YjgEeA851z5WbWGogCPgUGO+f2ehpQREQkQDNYIiLSELQH9jrnygECBdX3gBTgfTN7H8DMLjazRWa23MxeNLP4wPYtZva/Zva5mS0xs26B7Veb2WozW2lmH3nz1kREpDHRDJaIiIS8QKH0CRALLABmOec+NLMtBGawArNaLwOjnXNHzOxuINo590Bgv787535jZtcB451zl5nZ58Ao59wOM2vunDvgyRsUEZFGQzNYIiIS8pxzh4FBwBSgEJhlZjccs9twIANYaGYrgOuB9BqPv1Dj54jA7YXAM2b2QyC8btKLiEhTEuF1ABERkdpwzvmAD4APAjNP1x+ziwHvOOcmnugljr3tnLvFzIYB3wWWmdkg59y+4CYXEZGmRDNYIiIS8sysp5l1r7GpP5APFAMJgW3ZwDk1rq+KM7MeNZ5zTY2fiwL7dHXOLXbO/YrqmbEOdfg2RESkCdAMloiINATxwGNm1hyoAjZSvVxwIvCWmRU45y4ILBt8wcyiA8/7JbA+cLuFma0CygPPA/hDoHAz4F1gZb28GxERabTU5EJERBq9ms0wvM4iIiKNm5YIioiIiIiIBIlmsERERERERIJEM1giIiIiIiJBogJLREREREQkSFRgiYiIiIiIBIkKLBERERERkSBRgSUiIiIiIhIk/weiOP2N0fDDygAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 864x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Average accuracy: 0.8081352710723877\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PXAMLWLhNFcT",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "outputId": "cf48de96-91d8-42db-822e-3214652ebb69"
      },
      "source": [
        "cm_data = {'preds': preds, 'true': labels}\n",
        "cm_df = pd.DataFrame.from_dict(cm_data)\n",
        "cm = pd.crosstab(cm_df['true'], cm_df['preds'], rownames=['Actual'], colnames=['Predicted'])\n",
        "\n",
        "sn.heatmap(cm, annot=True)\n",
        "plt.show()"
      ],
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pjH0di1woTN8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def predict(text):\n",
        "    clean_text = clean(text)\n",
        "    tokenized = [word for word in clean_text.split(' ') if word != '']\n",
        "    indexed = [TEXT.vocab.stoi[w] for w in tokenized]\n",
        "    length = [len(indexed)]\n",
        "    tensor = torch.LongTensor(indexed).to(device)\n",
        "    tensor = tensor.unsqueeze(1)\n",
        "    length_tensor = torch.LongTensor(length)\n",
        "    prediction = torch.sigmoid(model(tensor, length_tensor))\n",
        "    return prediction.item()"
      ],
      "execution_count": 57,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9M-9Wl4fpSJd",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "2a53c9e2-8d15-483a-e2f1-66a4c57e6924"
      },
      "source": [
        "predict(\"i hate you\")"
      ],
      "execution_count": 58,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.05238809064030647"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 58
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OicO6TN2qMMi",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "960cc7a8-fd23-4867-ed40-3bb2fb72d112"
      },
      "source": [
        "predict(\"took my dog out for a walk, the weather was pretty good\")"
      ],
      "execution_count": 59,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.905511200428009"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 59
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PcHc5irEm2CA",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import pickle\n",
        "\n",
        "torch.save(model.state_dict(), './twitter_sentiment_state_dict.pt')\n",
        "\n",
        "model_config = {'stoi' : TEXT.vocab.stoi,\n",
        "        'input_dim': INPUT_DIM,\n",
        "        'hidden_dim': HIDDEN_DIM,\n",
        "        'embedding_dim': EMBEDDING_DIM,\n",
        "        'output_dim': OUTPUT_DIM,\n",
        "        'n_layers': N_LAYERS,\n",
        "        'pad_idx': PAD_IDX}\n",
        "\n",
        "with open('config.pickle', 'wb') as handle:\n",
        "    pickle.dump(model_config, handle, protocol=pickle.HIGHEST_PROTOCOL)"
      ],
      "execution_count": 36,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xeme2msGKDUG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}