NLP/NewsGroup_ULMFiT_fastai_Text_Classification.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "NewsGroup:ULMFiT_fastai_Text_Classification",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "metadata": {
        "id": "rsMgf8LLsoOq",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "##Install Dependencies\n",
        "Let's install PyTorch and fastai libraries first. You have to repeat this step every time you restart this notebook in colab."
      ]
    },
    {
      "metadata": {
        "id": "9L-mYNWDBB4g",
        "colab_type": "code",
        "outputId": "348cf3e3-abd6-4768-dee1-bfc77b212c94",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1431
        }
      },
      "cell_type": "code",
      "source": [
        "!pip install torch_nightly -f https://download.pytorch.org/whl/nightly/cu92/torch_nightly.html\n",
        "!pip install fastai"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Looking in links: https://download.pytorch.org/whl/nightly/cu92/torch_nightly.html\n",
            "Collecting torch_nightly\n",
            "\u001b[?25l  Downloading https://download.pytorch.org/whl/nightly/cu92/torch_nightly-1.0.0.dev20181206-cp36-cp36m-linux_x86_64.whl (576.2MB)\n",
            "\u001b[K    100% |████████████████████████████████| 576.2MB 26kB/s \n",
            "tcmalloc: large alloc 1073750016 bytes == 0x635d2000 @  0x7efe5d6042a4 0x591a07 0x5b5d56 0x502e9a 0x506859 0x502209 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x507641 0x502209 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x507641 0x504c28 0x502540 0x502f3d 0x507641\n",
            "\u001b[?25hInstalling collected packages: torch-nightly\n",
            "Successfully installed torch-nightly-1.0.0.dev20181206\n",
            "Collecting fastai\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/81/e6/e76ebda02f5aeb5c387ffc83174d9d0fa5a95b8b30f9020758e3c42da51b/fastai-1.0.38-py3-none-any.whl (148kB)\n",
            "\u001b[K    100% |████████████████████████████████| 153kB 6.2MB/s \n",
            "\u001b[?25hRequirement already satisfied: scipy in /usr/local/lib/python3.6/dist-packages (from fastai) (1.1.0)\n",
            "Requirement already satisfied: matplotlib in /usr/local/lib/python3.6/dist-packages (from fastai) (2.1.2)\n",
            "Collecting thinc==6.12.0 (from fastai)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/1f/18/e320bfc57c20df39cc5ffa1915c7b5402a9038f290ddd85b5b72689bd57a/thinc-6.12.0-cp36-cp36m-manylinux1_x86_64.whl (1.9MB)\n",
            "\u001b[K    100% |████████████████████████████████| 1.9MB 14.2MB/s \n",
            "\u001b[?25hCollecting nvidia-ml-py3 (from fastai)\n",
            "  Downloading https://files.pythonhosted.org/packages/6d/64/cce82bddb80c0b0f5c703bbdafa94bfb69a1c5ad7a79cff00b482468f0d3/nvidia-ml-py3-7.352.0.tar.gz\n",
            "Requirement already satisfied: pyyaml in /usr/local/lib/python3.6/dist-packages (from fastai) (3.13)\n",
            "Collecting torch (from fastai)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/7e/60/66415660aa46b23b5e1b72bc762e816736ce8d7260213e22365af51e8f9c/torch-1.0.0-cp36-cp36m-manylinux1_x86_64.whl (591.8MB)\n",
            "\u001b[K    100% |████████████████████████████████| 591.8MB 27kB/s \n",
            "tcmalloc: large alloc 1073750016 bytes == 0x609d6000 @  0x7fd9d75fa2a4 0x591a07 0x5b5d56 0x502e9a 0x506859 0x502209 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x507641 0x502209 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x507641 0x504c28 0x502540 0x502f3d 0x507641\n",
            "\u001b[?25hRequirement already satisfied: packaging in /usr/local/lib/python3.6/dist-packages (from fastai) (18.0)\n",
            "Collecting dataclasses; python_version < \"3.7\" (from fastai)\n",
            "  Downloading https://files.pythonhosted.org/packages/26/2f/1095cdc2868052dd1e64520f7c0d5c8c550ad297e944e641dbf1ffbb9a5d/dataclasses-0.6-py3-none-any.whl\n",
            "Requirement already satisfied: pandas in /usr/local/lib/python3.6/dist-packages (from fastai) (0.22.0)\n",
            "Collecting torchvision (from fastai)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/ca/0d/f00b2885711e08bd71242ebe7b96561e6f6d01fdb4b9dcf4d37e2e13c5e1/torchvision-0.2.1-py2.py3-none-any.whl (54kB)\n",
            "\u001b[K    100% |████████████████████████████████| 61kB 19.7MB/s \n",
            "\u001b[?25hRequirement already satisfied: cymem==2.0.2 in /usr/local/lib/python3.6/dist-packages (from fastai) (2.0.2)\n",
            "Requirement already satisfied: numpy>=1.12 in /usr/local/lib/python3.6/dist-packages (from fastai) (1.14.6)\n",
            "Collecting fastprogress>=0.1.18 (from fastai)\n",
            "  Downloading https://files.pythonhosted.org/packages/78/57/24a5e20f4a357f7f1c90dd5250071951c832b2480fd4fefd7be48edf4180/fastprogress-0.1.18-py3-none-any.whl\n",
            "Requirement already satisfied: Pillow in /usr/local/lib/python3.6/dist-packages (from fastai) (4.0.0)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from fastai) (2.18.4)\n",
            "Collecting spacy==2.0.16 (from fastai)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/ed/39/288640f591b29aac6996c97ddfafc3262ae0be7513e06bc560921b112d7c/spacy-2.0.16-cp36-cp36m-manylinux1_x86_64.whl (23.3MB)\n",
            "\u001b[K    100% |████████████████████████████████| 23.3MB 1.5MB/s \n",
            "\u001b[?25hRequirement already satisfied: numexpr in /usr/local/lib/python3.6/dist-packages (from fastai) (2.6.8)\n",
            "Requirement already satisfied: regex==2018.01.10 in /usr/local/lib/python3.6/dist-packages (from fastai) (2018.1.10)\n",
            "Collecting typing (from fastai)\n",
            "  Downloading https://files.pythonhosted.org/packages/4a/bd/eee1157fc2d8514970b345d69cb9975dcd1e42cd7e61146ed841f6e68309/typing-3.6.6-py3-none-any.whl\n",
            "Collecting bottleneck (from fastai)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/05/ae/cedf5323f398ab4e4ff92d6c431a3e1c6a186f9b41ab3e8258dff786a290/Bottleneck-1.2.1.tar.gz (105kB)\n",
            "\u001b[K    100% |████████████████████████████████| 112kB 28.1MB/s \n",
            "\u001b[?25hRequirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib->fastai) (0.10.0)\n",
            "Requirement already satisfied: pytz in /usr/local/lib/python3.6/dist-packages (from matplotlib->fastai) (2018.7)\n",
            "Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->fastai) (2.5.3)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->fastai) (2.3.0)\n",
            "Requirement already satisfied: six>=1.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib->fastai) (1.11.0)\n",
            "Requirement already satisfied: cytoolz<0.10,>=0.9.0 in /usr/local/lib/python3.6/dist-packages (from thinc==6.12.0->fastai) (0.9.0.1)\n",
            "Requirement already satisfied: plac<1.0.0,>=0.9.6 in /usr/local/lib/python3.6/dist-packages (from thinc==6.12.0->fastai) (0.9.6)\n",
            "Requirement already satisfied: dill<0.3.0,>=0.2.7 in /usr/local/lib/python3.6/dist-packages (from thinc==6.12.0->fastai) (0.2.8.2)\n",
            "Requirement already satisfied: msgpack<1.0.0,>=0.5.6 in /usr/local/lib/python3.6/dist-packages (from thinc==6.12.0->fastai) (0.5.6)\n",
            "Requirement already satisfied: msgpack-numpy<0.4.4.0 in /usr/local/lib/python3.6/dist-packages (from thinc==6.12.0->fastai) (0.4.3.2)\n",
            "Requirement already satisfied: tqdm<5.0.0,>=4.10.0 in /usr/local/lib/python3.6/dist-packages (from thinc==6.12.0->fastai) (4.28.1)\n",
            "Requirement already satisfied: preshed<3.0.0,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from thinc==6.12.0->fastai) (2.0.1)\n",
            "Requirement already satisfied: wrapt<1.11.0,>=1.10.0 in /usr/local/lib/python3.6/dist-packages (from thinc==6.12.0->fastai) (1.10.11)\n",
            "Requirement already satisfied: murmurhash<1.1.0,>=0.28.0 in /usr/local/lib/python3.6/dist-packages (from thinc==6.12.0->fastai) (1.0.1)\n",
            "Requirement already satisfied: olefile in /usr/local/lib/python3.6/dist-packages (from Pillow->fastai) (0.46)\n",
            "Requirement already satisfied: idna<2.7,>=2.5 in /usr/local/lib/python3.6/dist-packages (from requests->fastai) (2.6)\n",
            "Requirement already satisfied: chardet<3.1.0,>=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests->fastai) (3.0.4)\n",
            "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests->fastai) (2018.11.29)\n",
            "Requirement already satisfied: urllib3<1.23,>=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests->fastai) (1.22)\n",
            "Requirement already satisfied: ujson>=1.35 in /usr/local/lib/python3.6/dist-packages (from spacy==2.0.16->fastai) (1.35)\n",
            "Requirement already satisfied: toolz>=0.8.0 in /usr/local/lib/python3.6/dist-packages (from cytoolz<0.10,>=0.9.0->thinc==6.12.0->fastai) (0.9.0)\n",
            "Building wheels for collected packages: nvidia-ml-py3, bottleneck\n",
            "  Running setup.py bdist_wheel for nvidia-ml-py3 ... \u001b[?25l-\b \bdone\n",
            "\u001b[?25h  Stored in directory: /root/.cache/pip/wheels/e4/1d/06/640c93f5270d67d0247f30be91f232700d19023f9e66d735c7\n",
            "  Running setup.py bdist_wheel for bottleneck ... \u001b[?25l-\b \b\\\b \b|\b \b/\b \b-\b \b\\\b \b|\b \b/\b \bdone\n",
            "\u001b[?25h  Stored in directory: /root/.cache/pip/wheels/f2/bf/ec/e0f39aa27001525ad455139ee57ec7d0776fe074dfd78c97e4\n",
            "Successfully built nvidia-ml-py3 bottleneck\n",
            "\u001b[31mspacy 2.0.16 has requirement numpy>=1.15.0, but you'll have numpy 1.14.6 which is incompatible.\u001b[0m\n",
            "\u001b[31mtorchvision 0.2.1 has requirement pillow>=4.1.1, but you'll have pillow 4.0.0 which is incompatible.\u001b[0m\n",
            "Installing collected packages: thinc, nvidia-ml-py3, torch, dataclasses, torchvision, fastprogress, spacy, typing, bottleneck, fastai\n",
            "  Found existing installation: thinc 6.12.1\n",
            "    Uninstalling thinc-6.12.1:\n",
            "      Successfully uninstalled thinc-6.12.1\n",
            "  Found existing installation: spacy 2.0.18\n",
            "    Uninstalling spacy-2.0.18:\n",
            "      Successfully uninstalled spacy-2.0.18\n",
            "Successfully installed bottleneck-1.2.1 dataclasses-0.6 fastai-1.0.38 fastprogress-0.1.18 nvidia-ml-py3-7.352.0 spacy-2.0.16 thinc-6.12.0 torch-1.0.0 torchvision-0.2.1 typing-3.6.6\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "SA87AbrrBigr",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# import libraries\n",
        "import fastai\n",
        "from fastai import *\n",
        "from fastai.text import * \n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "from functools import partial\n",
        "import io\n",
        "import os"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "1rM5TMVW3AQ-",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Import Dataset"
      ]
    },
    {
      "metadata": {
        "id": "r9SHCfiABlcF",
        "colab_type": "code",
        "outputId": "3bf53644-f918-4ea2-80e5-54824cfc600c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "cell_type": "code",
      "source": [
        "#import dataset\n",
        "from sklearn.datasets import fetch_20newsgroups\n",
        "dataset = fetch_20newsgroups(shuffle=True, random_state=1, remove=('headers', 'footers', 'quotes'))\n",
        "documents = dataset.data"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading 20news dataset. This may take a few minutes.\n",
            "Downloading dataset from https://ndownloader.figshare.com/files/5975967 (14 MB)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "metadata": {
        "id": "Xvs8MygLEKri",
        "colab_type": "code",
        "outputId": "891e22b4-dc54-4f4b-97c3-376313fed95a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 357
        }
      },
      "cell_type": "code",
      "source": [
        "dataset.target_names"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['alt.atheism',\n",
              " 'comp.graphics',\n",
              " 'comp.os.ms-windows.misc',\n",
              " 'comp.sys.ibm.pc.hardware',\n",
              " 'comp.sys.mac.hardware',\n",
              " 'comp.windows.x',\n",
              " 'misc.forsale',\n",
              " 'rec.autos',\n",
              " 'rec.motorcycles',\n",
              " 'rec.sport.baseball',\n",
              " 'rec.sport.hockey',\n",
              " 'sci.crypt',\n",
              " 'sci.electronics',\n",
              " 'sci.med',\n",
              " 'sci.space',\n",
              " 'soc.religion.christian',\n",
              " 'talk.politics.guns',\n",
              " 'talk.politics.mideast',\n",
              " 'talk.politics.misc',\n",
              " 'talk.religion.misc']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "metadata": {
        "id": "2OWzByac3GXG",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "##Create Dataframe\n",
        "This provides the Label/Target and the associated text"
      ]
    },
    {
      "metadata": {
        "id": "A3zbA9rfCfRh",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# create a dataframe\n",
        "df = pd.DataFrame({'label':dataset.target,\n",
        "                   'text':dataset.data})"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "lifPA3p6C853",
        "colab_type": "code",
        "outputId": "47de8b46-2cb2-4715-c017-e572eb43435a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "df.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(11314, 2)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "metadata": {
        "id": "_bk9iNNh3QEH",
        "colab_type": "code",
        "outputId": "39f5fb9e-3c44-44d5-a125-cb99ce1506e6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "cell_type": "code",
      "source": [
        "df.head()"
      ],
      "execution_count": 0,
      "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>label</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>17</td>\n",
              "      <td>Well i'm not sure about the story nad it did s...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0</td>\n",
              "      <td>\\n\\n\\n\\n\\n\\n\\nYeah, do you expect people to re...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>17</td>\n",
              "      <td>Although I realize that principle is not one o...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>11</td>\n",
              "      <td>Notwithstanding all the legitimate fuss about ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>10</td>\n",
              "      <td>Well, I will have to change the scoring on my ...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   label                                               text\n",
              "0     17  Well i'm not sure about the story nad it did s...\n",
              "1      0  \\n\\n\\n\\n\\n\\n\\nYeah, do you expect people to re...\n",
              "2     17  Although I realize that principle is not one o...\n",
              "3     11  Notwithstanding all the legitimate fuss about ...\n",
              "4     10  Well, I will have to change the scoring on my ..."
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "metadata": {
        "id": "1P5OelaiCmna",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "df = df[df['label'].isin([1,10])]\n",
        "df = df.reset_index(drop = True)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "6zbrKEr7C419",
        "colab_type": "code",
        "outputId": "76aa947e-a64e-4c2a-f4e7-83011b7ad5da",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "cell_type": "code",
      "source": [
        "df['label'].value_counts()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "10    600\n",
              "1     584\n",
              "Name: label, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "metadata": {
        "id": "_SXRszhllc-E",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Preprocessing\n",
        "\n",
        "Tokenization\n",
        "\n",
        "Remove StopWords"
      ]
    },
    {
      "metadata": {
        "id": "THgv8hHrC8wD",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "df['text'] = df['text'].str.replace(\"[^a-zA-Z]\", \" \")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "Uxc4Yd76Dw73",
        "colab_type": "code",
        "outputId": "7ed78ea8-28cc-4514-a4c6-d2655b523ec7",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "cell_type": "code",
      "source": [
        "import nltk\n",
        "nltk.download('stopwords')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Unzipping corpora/stopwords.zip.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "True"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "metadata": {
        "id": "IqG4He_qD1HW",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from nltk.corpus import stopwords \n",
        "stop_words = stopwords.words('english') "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "UmZQe0z9D31M",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# tokenization \n",
        "tokenized_doc = df['text'].apply(lambda x: x.split())\n",
        "\n",
        "# remove stop-words \n",
        "tokenized_doc = tokenized_doc.apply(lambda x: [item for item in x if item not in stop_words]) \n",
        "\n",
        "# de-tokenization \n",
        "detokenized_doc = [] \n",
        "for i in range(len(df)): \n",
        "    t = ' '.join(tokenized_doc[i]) \n",
        "    detokenized_doc.append(t) \n",
        "df['text'] = detokenized_doc"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "MK_i0X2kD6AC",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "# split data into training and validation set\n",
        "df_trn, df_val = train_test_split(df, stratify = df['label'], test_size = 0.4, random_state = 12)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "IYEFCrJOEHSG",
        "colab_type": "code",
        "outputId": "cc2d73bc-21ea-4f57-c55c-44dd5de5863a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "df_trn.shape, df_val.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((710, 2), (474, 2))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "metadata": {
        "id": "xvvFWWM4ljoH",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "## Data Preparation\n",
        "\n",
        "### Language model data\n",
        "for learning the Language semantics\n",
        "\n",
        "### Classifier model data\n",
        "for learning the classification Task on text data\n",
        "\n",
        "**text.learner contains helper functions to quickly create a language model or an RNN classifier**\n"
      ]
    },
    {
      "metadata": {
        "id": "7tNP_aMTEOKe",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Language model data\n",
        "data_lm = TextLMDataBunch.from_df(train_df = df_trn, valid_df = df_val, path = \"\")\n",
        "\n",
        "# Classifier model data\n",
        "data_clas = TextClasDataBunch.from_df(path = \"\", train_df = df_trn, valid_df = df_val, vocab=data_lm.train_ds.vocab, bs=32)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "LoQ60UxB49E-",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "This does all the necessary preprocessing behing the scene. For the classifier, we also pass the vocabulary (mapping from ids to words) that we want to use: this is to ensure that data_clas will use the same dictionary as data_lm.\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "8Ua8pmHEEP5N",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "#Fine-tuning a language model\n",
        "#We can use the data_lm object we created earlier to fine-tune a pretrained language model.\n",
        "#fast.ai has an English model available that we can download. \n",
        "#We can create a learner object that will directly create a model, \n",
        "#download the pretrained weights and be ready for fine-tuning.\n",
        "learn = language_model_learner(data_lm, pretrained_model=URLs.WT103, drop_mult=0.7)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "YE3likAPETlz",
        "colab_type": "code",
        "outputId": "7337eee7-ab7e-4aa3-a423-8ee45562f526",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 94
        }
      },
      "cell_type": "code",
      "source": [
        "# train the learner object\n",
        "learn.fit_one_cycle(1, 1e-2)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "Total time: 00:09 <p><table style='width:300px; margin-bottom:10px'>\n",
              "  <tr>\n",
              "    <th>epoch</th>\n",
              "    <th>train_loss</th>\n",
              "    <th>valid_loss</th>\n",
              "    <th>accuracy</th>\n",
              "  </tr>\n",
              "  <tr>\n",
              "    <th>1</th>\n",
              "    <th>7.808674</th>\n",
              "    <th>6.311074</th>\n",
              "    <th>0.129677</th>\n",
              "  </tr>\n",
              "</table>\n"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "w3eJMLWDKb9u",
        "colab_type": "code",
        "outputId": "e464881c-38bc-4c3e-b79d-b4c1fe04a65f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 170
        }
      },
      "cell_type": "code",
      "source": [
        "learn.lr_find()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='3' class='' max='4', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      75.00% [3/4 00:21<00:07]\n",
              "    </div>\n",
              "    \n",
              "<table style='width:300px; margin-bottom:10px'>\n",
              "  <tr>\n",
              "    <th>epoch</th>\n",
              "    <th>train_loss</th>\n",
              "    <th>valid_loss</th>\n",
              "    <th>accuracy</th>\n",
              "  </tr>\n",
              "  <tr>\n",
              "    <th>1</th>\n",
              "    <th>6.922407</th>\n",
              "  </tr>\n",
              "  <tr>\n",
              "    <th>2</th>\n",
              "    <th>6.910876</th>\n",
              "  </tr>\n",
              "  <tr>\n",
              "    <th>3</th>\n",
              "    <th>6.692184</th>\n",
              "  </tr>\n",
              "</table>\n",
              "\n",
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='0' class='progress-bar-interrupted' max='25', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      Interrupted\n",
              "    </div>\n",
              "    "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "81fA6MpQKpJw",
        "colab_type": "code",
        "outputId": "90cf5b66-2cff-42da-f9f9-dd9f0203776a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 361
        }
      },
      "cell_type": "code",
      "source": [
        "learn.recorder.plot()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFYCAYAAAB6RnQAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzt3Xl41OW9///nLJnsO5OQhATCjqxh\nESngjksXK5SAeFw49adH5WhbbatHofIt9rRal7pQBbXV4lILdYlHsB6t2GohyCIhQAh7yD4heybJ\nZDLz+yNmlMOWhExmJvN6XBeXzExm5v1OMK/53J/7c98Gt9vtRkRERAKG0dcFiIiISPcovEVERAKM\nwltERCTAKLxFREQCjMJbREQkwCi8RUREAozZ1wV0lc3W4OsSiI+PoKbG7usy+oz67f+Credg6xeC\nr+f+1q/VGn3K+3Xk3Q1ms8nXJfQp9dv/BVvPwdYvBF/PwdKvwltERCTAKLxFREQCjMJbREQkwCi8\nRUREAozCW0REJMAovEVERAKMwltERCTAKLxFREQCjMJbREQkwCi8RUREAozCW0RE5BzVNraSu6cC\nt9vdJ++n8BYRETlHf9tSxKqc3dhqm/vk/RTeIiIi56iuyQGAydg3sarwFhEROUf2FicAEWF9s9O2\nwltEROQcNbW0YTQYCLP0zZakCm8REZFzZG9xEhFmxmAw9Mn7KbxFRETOUWd49xWFt4iIyDlqanES\nqfAWEREJDI62dpztLiLCQvrsPRXeIiIi58De2jHTXEfeIiIiAaKp8zKx0L4Lb6+9U1NTE/fddx91\ndXW0tbWxZMkSZs+e7Xk8JyeHV155BaPRyIIFC8jOzvZWKSIiIl5jb2kD6NNhc6+F99tvv01mZib3\n3nsvFRUV3HzzzXzwwQcA2O12Vq5cybp16wgJCWH+/PnMmTOHuLg4b5UjIiLiFZ1H3v1i2Dw+Pp7a\n2loA6uvriY+P9zy2c+dOxo8fT3R0NGFhYUyePJnt27d7qxQRERGvae7j1dXAi0fe3/nOd3jrrbeY\nM2cO9fX1rFq1yvNYVVUVCQkJntsJCQnYbDZvlSIiIuI1Tf1p2Pzdd98lNTWVl156iYKCAh544AHe\neuutU35tV7ZQi4+PwGzum2XnzsRqjfZ1CX1K/fZ/wdZzsPULwddzn/dr6sim1OToPntvr4X39u3b\nmTVrFgCjR4+msrKS9vZ2TCYTSUlJVFVVeb62srKSSZMmnfH1amrs3iq1y6zWaGy2Bl+X0WfUb/8X\nbD0HW78QfD37ol/b8SYA2lrbev29T/dhwGvnvAcPHszOnTsBKCkpITIyEtNXn04mTpzIrl27qK+v\np6mpie3btzN16lRvlSIiIuI19tZ+NGy+cOFCHnjgAW644QacTifLly9n9erVTJs2jaysLO69915u\nueUWDAYDS5YsITo6uIZ1RESkf7D3p+u8IyMjeeqpp064b8aMGZ6/X3XVVVx11VXeensREZE+4YtF\nWrTCmoiIyDmwt7QRHmrGaOyb7UBB4S0iInJO7K19u6MYKLxFRETOSVOLs0+HzEHhLSIi0mPOdhet\njvY+XV0NFN4iIiI99vV2oH13mRgovEVERHrMF+uag8JbRESkx5oU3iIiIoHFF3t5g8JbRESkx3yx\nlzcovEVERHqsc8Kahs1FREQChGfYPFTD5iIiIgFBw+YiIiIB5usJawpvERGRgGBv0SItIiIiAUXX\neYuIiAQYe4sTS4gRs6lv41ThLSIi0kNNLW19PmQOCm8REZEea2519vmQOSi8RUREesTldmP3wV7e\noPAWERHpkZZWJ276fqY5KLxFRER6xFczzUHhLSIi0iN2hbeIiEhg+Xpdc4W3iIhIQGjy0epqoPAW\nERHpEV9tBwoKbxERkR7x1brmoPAWERHpkSYf7SgG4LV3XLt2LTk5OZ7b+fn57Nixw3P7ySefJDc3\nF7fbzeWXX86tt97qrVJERER6nS9nm3vtHbOzs8nOzgZgy5YtbNiwwfNYYWEhubm5/PnPf8blcvGd\n73yHa6+9FqvV6q1yREREelXnkXe/HTZfuXIld955p+d2dHQ0ra2tOBwOWltbMRqNhIeH90UpIiIi\nvcKXE9a8/o55eXmkpKSccFSdkpLCVVddxSWXXEJ7eztLliwhKirK26WIiIj0GnuLE5PRgMXc99PH\nvB7e69atY+7cuSfcd+zYMf73f/+Xjz76CKfTyXXXXce3v/1tEhMTT/s68fERmM0mb5d7VlZrtK9L\n6FPqt/8Ltp6DrV8Ivp77qt/WtnaiIywkJcX0yft9k9fDOzc3l6VLl55w365du5g4caJnqHzUqFEU\nFhYyY8aM075OTY3dq3V2hdUajc3W4Osy+oz67f+Credg6xeCr+e+7Le+yUFkWIhX3+90H0S8eqxf\nUVFBZGQkFovlhPszMjLIz8/H5XLR1tZGYWEh6enp3ixFRESk17i/2g400gfnu8HLR942m42EhATP\n7dWrVzNt2jSysrKYOXMm119/PQDz589n0KBB3ixFRESk1zjaXLS73IT3x/AeN24cL774ouf2bbfd\n5vn73Xffzd133+3NtxcREfEKX14mBlphTUREpNt8uUALKLxFRES6zXONtw+2AwWFt4iISLdp2FxE\nRCTAaNhcREQkwDR5tgNVeIuIiAQEe+d2oDrnLSIiEhi+HjbXOW8REZGAoGFzERGRAOMZNteRt4iI\nSGCwtzoxAGGhvtntUuEtIiLSTfYWJxFhZowGg0/eX+EtIiLSTU0tbT67xhsU3iIiIt3WceTtm/Pd\noPAWERHpljanC4fT5bNrvEHhLSIi0i2dm5L46jIxUHiLiIh0i68vEwOFt4iISLf4eoEWUHiLiIh0\ni693FAOFt4iISLdo2FxERCTANDZ3hLeGzUVERAJEXZMDgNhIi89qUHiLiIh0gye8o0J9VoPCW0RE\npBvqdeQtIiISWOqaHFjMRsIsvtlRDBTeIiIi3VLf5CAm0oLBRzuKgcJbRESky1xuN/VNDmKjfDdk\nDgpvERGRLmtqbqPd5SY20neT1QC8dpHa2rVrycnJ8dzOz89nx44dntsFBQU88MADAFx22WUsWbLE\nW6WIiIj0is6Z5jE+nKwGXgzv7OxssrOzAdiyZQsbNmw44fFly5axYsUKxowZw09/+lOam5sJDw/3\nVjkiIiLnzB+u8QYvhvc3rVy5kscee8xzu6qqCrvdztixYwF44okn+qIMERGRc+IPl4lBH5zzzsvL\nIyUlBavV6rmvpKSE2NhY7r//fq677jpefvllb5chIiJyzuoa/SO8vX7kvW7dOubOnXvCfW63m+Li\nYlauXElYWBgLFy5k5syZjBgx4rSvEx8fgdnsu2vqOlmt0b4uoU+p3/4v2HoOtn4h+Hr2Zr9t7o7/\nDh4U59Pvq9fDOzc3l6VLl55wX2JiIiNGjCA+Ph6AKVOmsH///jOGd02N3at1doXVGo3N1uDrMvqM\n+u3/gq3nYOsXgq9nb/db/tVru9qcffJ9Pd0HBK8Om1dUVBAZGYnFcuLwQnp6Ok1NTdTW1uJyudi7\ndy9Dhw71ZikiIiLnLCgmrNlsNhISEjy3V69ezbRp08jKyuK//uu/uPXWWzEYDMyePZvRo0d7sxQR\nEZFzVtfkICLUTIiPT+N6NbzHjRvHiy++6Ll92223ef4+ceJE1q5d6823FxER6VV1jQ6fX+MNWmFN\nRESkS5ztLpqa23w+ZA4KbxERkS5psLfhBp+vaw4KbxERkS6p95OlUUHhLSIi0iV1Ta2A72eag8Jb\nRESkS75eXc23O4qBwltERKRLPNd465y3iIhIYPCXBVpA4S0iItIlmrAmIiISYOqaHBiA6IgQX5ei\n8BYREemKuiYH0REhmIy+j07fVyAiIhIA6ptaifGDmeag8BYRETmr1rZ2mlvb/WKmOSi8RUREzqre\nj2aag8JbRETkrPzpMjFQeIuIiJyVjrxFREQCTJ0fXeMNCm8REZGzqmv0n01JQOEtIiJyVp7V1aJ0\nqZiIiEhA0IQ1ERGRAFPX5MBkNBAZZvZ1KYDCW0RE5KzqGh3ERlkwGAy+LgVQeIuIiJyR2+2m3u7w\nmyFzUHiLiIicUXNrO21OF7F+sq45KLxFRETOqK6p4zKxmEjfbwXaSeEtIiJyBp7LxHTkLSIiEhj8\n7TIxUHiLiIicUV2j/4W31y5YW7t2LTk5OZ7b+fn57Nix46Svu+eee7BYLPzmN7/xVikiIiI95jny\n9pO9vMGL4Z2dnU12djYAW7ZsYcOGDSd9zeeff05RURHDhw/3VhkiIiLnxN92FIM+GjZfuXIld955\n5wn3ORwOnnvuOe64446+KEFERKRH/G1HMeiD8M7LyyMlJQWr1XrC/atWrWLRokVERUV5uwQREZEe\nq2tqJdRiIsziH0ujgheHzTutW7eOuXPnnnDfkSNHyM/P56677iI3N7dLrxMfH4HZbPJGid1itUb7\nuoQ+pX77v2DrOdj6heDrubf7bWxuIyE6zK++j14P79zcXJYuXXrCfRs3bqS0tJQFCxbQ2NhIdXU1\nL7zwArfeeutpX6emxu7tUs/Kao3GZmvwdRl9Rv32f8HWc7D1C8HXc2/363K5qW1wMDQtzCffx9N9\nYPBqeFdUVBAZGYnFcuJ5gsWLF7N48WKgI9zffvvtMwa3iIiILzQ2t+Fyu/1qshp4+Zy3zWYjISHB\nc3v16tWnvFxMRETEH/njAi3g5SPvcePG8eKLL3pu33bbbSd9zfTp05k+fbo3yxAREemR2saOdc1j\no/xnaVTo4pF3fn4+n3zyCQBPPvkkN998M1u3bvVqYSIiIr52vL4FgAExYT6u5ERdCu+HH36YzMxM\ntm7dyq5du1i2bBlPP/20t2sTERHxqeqvwjshJgCPvENDQxkyZAgff/wxCxYsYPjw4RiNWhZdRET6\nt+N1HcPmCYF45N3c3MyGDRv46KOPmDVrFrW1tdTX13u7NhEREZ+qrm/BAMRHB+CR9z333MN7773H\nT37yE6KiolizZo3nUi8REZH+qrqhhdgoC2aTf402d2m2+QUXXMC4ceOIioqiqqqKGTNmMHnyZG/X\nJiIi4jMut5vq+laGDPSfldU6demjxIoVK9iwYQO1tbVcd911vPrqqyxfvtzLpYmIiPhOfZODdpeb\neD873w1dDO89e/aQnZ3Nhg0bmDt3Lr/73e84evSot2sTERHxmc7LxBL9bKY5dDG83W430LEm+aWX\nXgp0bOkpIiLSX1XX++dMc+hieGdmZvLtb3+bpqYmxowZwzvvvENsbKy3axMREfGZ43WdR97+F95d\nmrD28MMPU1hYyLBhwwAYPnw4jz76qFcLExER8aXqhgAP75aWFv7+97/z1FNPYTAYmDRpEsOHD/d2\nbSIiIj7TOWweH6jnvJctW0ZjYyPXXXcdCxYsoKqq6qQ9ukVERPqT4/UthJiNRIeH+LqUk3TpyLuq\nqoonnnjCc/uSSy7hxhtv9FpRIiIivlZd30JCTBgGg8HXpZyky8ujNjc3e27b7XZaW1u9VpSIiIgv\nOdraabC3+eVlYtDFI++FCxdy9dVXM27cOAB2797Nj370I68WJiIi4is1Df57mRh0Mbznz5/PzJkz\n2b17NwaDgWXLlrFmzRpv1yYiIuITnQu0JPjZhiSduhTeACkpKaSkpHhu5+XleaUgERERX/t6dTX/\nPPLu8TYpnauuiYiI9Dee1dVi+1l4++PsOxERkd7g70feZxw2v+iii04Z0m63m5qaGq8VJSIi4kvV\ngXzO+/XXX++rOkRERPxGdX0rUeEhWEJMvi7llM4Y3mlpaX1Vh4iIiF9wu91U17eQkhjp61JOq8fn\nvEVERPqjxuY2HE4XCX66QAsovEVERE7QOdPcXyergcJbRETkBJ4FWhTeIiIigcEz09yPh827vMJa\nd61du5acnBzP7fz8fHbs2OG5vX79ev7whz9gNBqZMWMGP/nJT7xVioiISJcFwrC518I7Ozub7Oxs\nALZs2cKGDRs8jzU3N/PYY4+Rk5NDZGQkCxYs4Hvf+x7Dhw/3VjkiIiJdomHzr6xcuZI777zTczs8\nPJycnByioqIwGAzExcVRW1vbF6WIiIicUXV9Cyajgdgoi69LOS2vh3deXh4pKSlYrdYT7o+KigJg\n3759lJSUMHHiRG+XIiIiclbH61uIjw7F6MfLgHtt2LzTunXrmDt37ikfO3LkCD/96U95/PHHCQkJ\nOePrxMdHYDb7fqUbqzXa1yX0KfXb/wVbz8HWLwRfz+fSb5vTRV2Tg/MyE/36++b18M7NzWXp0qUn\n3V9eXs6SJUt49NFHGTNmzFlfp6bG7o3yusVqjcZma/B1GX1G/fZ/wdZzsPULwdfzufZbVduM2w0x\n4Wa/+L6d7gOEV4fNKyoqiIyMxGI5+bzBgw8+yPLlyxk7dqw3SxAREemyQJisBl4+8rbZbCQkJHhu\nr169mmnTphEXF8fWrVt5+umnPY8tXryYyy67zJvliIiInFEgXCYGXg7vcePG8eKLL3pu33bbbZ6/\n79y505tvLSIi0m2BcuStFdZERES+Egirq4HCW0RExKO6ITCGzRXeIiIiXzle30J4qJnwUK9fjHVO\nFN4iIiJfqa5vIdHPh8xB4S0iIgJAU0sbza3tfj9ZDRTeIiIiAJRXdywGlhwf4eNKzk7hLSIiApQf\n7wjvgQnhPq7k7BTeIiIifH3kPTBBR94iIiIBwRPeiZE+ruTsFN4iIiJARbWdUIuJOD/ex7uTwltE\nRIKey+2moqaZgfERGPx4H+9OCm8REQl61XUttDldDEz0//PdoPAWEREJqMlqoPAWERGhTOEtIiIS\nWHTkLSIiEmC+XqBF4S0iIhIQyqvtxEeHEmox+bqULlF4i4hIUGt1tFPT0BowR92g8BYRkSD39cpq\nCm8REZGAEGiT1UDhLSIiQa4zvFMU3iIiIoFBR94BoriykeV/2MLOA1W+LkVERHys/Lgds8lIQkyY\nr0vpsqAMb5PJQHm1nWff2sX2QpuvyxERER9xu92U19hJTgjHaPT/DUk6BWV4pyRG8pMFEzGbjDz3\nTj5bCyp9XZKIiPhAbaODVkd7QA2ZQ5CGN8CojPiOADcbef7d3WzZW+HrkkREpI+VH28CAut8NwRx\neAOMTI/j3oWTsIQYWZWzm827y31dkoiI9KFAnKwGYPbWC69du5acnBzP7fz8fHbs2OG5nZOTwyuv\nvILRaGTBggVkZ2d7q5QzGp4Wy73XTeKJN3fywnt7KDxWy7yLhhEVHuKTekREpO+UBeACLeDF8M7O\nzvYE8pYtW9iwYYPnMbvdzsqVK1m3bh0hISHMnz+fOXPmEBcX561yzmhYaiw/X5TFC/+zh41flrJ1\nn40Flwxn5viBGAyBM4FBRES6JxCv8YY+GjZfuXIld955p+f2zp07GT9+PNHR0YSFhTF58mS2b9/e\nF6Wc1uCB0Sz/92lkXzIMh7OdP6zfy29e286xykaf1iUiIt5TftxOTEQIEWGBNdrqtSPvTnl5eaSk\npGC1Wj33VVVVkZCQ4LmdkJCAzeb7S7bMJiNXTx/M+aOT+fPH+9lWaGP5H7YweaSVqy8YjNUa7esS\nRUSkl7Q52zle18KIQbG+LqXbvB7e69atY+7cuWf8GrfbfdbXiY+PwGzum63arNZolg+3sq2ggtc+\nKGBboY1thTYmDD/CDy4dQdZIa9AMpwfbB5Zg6xeCr+dg6xeCr+eu9nu0rB43MCQtLuC+R14P79zc\nXJYuXXrCfUlJSVRVfb26WWVlJZMmTTrj69TU2L1S35lkJEZw//VZFBTVsn7zUfIOVJF3oIqkuHDG\nDIlnzOB4RmfEExNp6fPa+oLVGo3N1uDrMvpMsPULwddzsPULwddzd/rdc6BjxDcuIsRvv0en+1Dh\n1fCuqKggMjISi+XEcJs4cSJLly6lvr4ek8nE9u3beeCBB7xZSo8ZDAbGDO4I6rrWdt74YC95B4/z\n6ZelfPplKQBpAyKZPSGFS6cMwmwK6qvvREQCRqBeJgZeDm+bzXbCue3Vq1czbdo0srKyuPfee7nl\nllswGAwsWbKE6Gj/H7IYPiiO278/jnaXiyPlDRQcraHgaA37i+v4898P8M9dZdx4xShGpvtm1ryI\niHRdIO7j3cng7soJZz/gD0MapxuOaWxuY93Gg/xjZ8eR+MxxA8m+ZHjAD6efbfipvslBQVENLY52\nXG43LlfHn3aXG4fThaOtHUebizZnOy2Odhqb22hobqOpuY3G5jbcboiLshAXFUpcdChxURZMRiPN\nDictrU6aW9tpbWsnzGIiITqMhJhQ4qM7/ljjwomLCu3VtYiDbXgRgq/nYOsXgq/n7vT7qz9t5Uh5\nA8/de5Hfjpr6ZNg8WESFh7D46tHMnpDCmg/38Xl+OTv2VzF1dBKjMuIYlR4XULvVnI7b7aa0qokv\nD1Tx5YEqDpV0TPbojhCzkajwEJLiwgGobXJQeay2268DYDIaGBAbxoC4cBJjwogKDyEizExEqJmI\nMDOhISb+77xCo9FAiMmIJcREiMlISIiRcEvH14tI8HC53ZQebyIpPtxvg/tM9BurFw1Li+UXN0/j\nkx0lvPPPQ/xjZ6nnaHxAbBijB8dz0cRUhqUFzmUJNQ2tFBR1nB7Ye7SGqroWAAwGGJEex8RhicRG\nWTAaDBiNBowGAyajgZAQIxazidAQEyFmI2EWE5HhIYSGnHzFgLPdRX2Tg5rGVtwuCAs1EW4xEx5q\nItRiwt7ipKahler6VmoaWjhe30pVXTNVdS3YapvZfbi6V3oNDzUREWomMiyEjORoxmYmMDYzQavt\nifRDlTXNNLe2M3G4/5+yPRWFdy8zGg1cNmUQF2elUlTRyL6iWgqP1bK/uJbP8sr4LK+MkYNiueqC\nwUwYlojRDy85c7vdfJBbxKY9FRR/Y5GaiFAzU0cnkTV8AOOHJfZaqHXuo3u60YnoCAvRERYykk/9\nP1mLw8nx+laaW5zYW9uwtzixtzppdbSf9LVOl5s2Z8dQvsPpoq3NRbPDib3FSWubi7rGFspr7BRV\nNvLZrjIMdCzgMzYzgYzkaJLjw0mOjyDU0jeXLYqIdxwpqwdgyGl+r/g7hbeXmIxGMlNiyEyJ4arp\nGbjcbvYV1fJBbhG7Dh2ncF0eaQMiuWJaOlNHJxEe6j8/ivWbj/LXTw8RZjExfmiiZ7Z9elKUX+53\nG2Yxkzbg3L9/nefKXG43RRUN5B+qZvfhag6U1HGk/MRzaPHRoSTEhOJ0umlxOGlxdJzXb3O6MBo5\nYRQiPNRMUnw4SfERJMWFkxwfTpo1EmtceNCsFyDibzr/nx6SEuPjSnrGfxKjnzN+45KzY5WNfJBb\nxJa9FfxxQwFrPixk/NAEpo1JYuKwAT4N8s93lfHXTw+RGBPK4z++CJfD6bNafMVoMDBkYAxDBsbw\n3W8NobnVyYGSOsqO26mosVNR3fHnUGk9FrOJMEvHn7ioUELMRtzujkl7LpebdrebRnsbe47UsOdI\nzQnvExUewrDUGIamxjA0LZZhqTGEWfS/pEhfOFLegAHISI7ydSk9ot8UPpCeFMWt3zuPeRcO5bNd\nZWwtqGTH/ip27K8ixGxkwtBEpp+XzIRhiVhOcY7YW/IPH+flDQVEhpn5yYJJJMaGB9Us1dMJDzUz\nfmgi44cmnnC/2+3u8pFzq6MdW20zFTXNVNbYOVrRwKHSenYePM7Og8eBjgl4Q1KiGZ3RsfjP8EGx\np5wj4C31dgflx+2YjIavPpCYCQvt+GBiMgbehB6R03G53RytaCBlQGTAfmAOzKr7icTYML4/K5Pv\nz8qkxNbIFwWVfFFQ6VmONdRiYvKIAUw/L5nzhiR4dUbk0fIGVr6dj8Fg4K4fTCB1QKTX3qu/6M6Q\nd6jFxKCkKAYlnfgpv66xlUOl9RworWNfUS2HSxs4WFLP+5uOYjIaGDEolvFDExk3NJFB1sheG2Z3\nud18ub+K0s1F7DtaTXFlI3VNjlN+rdFgICUxgvTkKDKSor/6bxTREYF9KaQEr4pqO62OdgYH6Plu\nUHj7jTRrFGnWKL4/K5NiWxO5eyrYsreCTbs7/ljMRoYMjPYMrw5NjSUuynLSL3OX2011XQslVU2U\nVjVR3dBKZko0YwYnEB8detL7ut1uSmxN/G7tThyOdu64dpwWmelDsVGhZI20kjWyY+Oe5lYn+4tr\nKSiqZe+RGgqKOv6+duNB4qIsjBmcwIDYMOKiLMRGhRIbZcEaG96tNQWqapv544YC9h79ehg/MSaM\nScMHkDIgAtx8dQ6/41x+vd1Bsa2JkqomNu+u8DwnPjqUjKQo0pOjyUiKYmhqTL+4JFL6vyNlnee7\nFd7SSwwGA+lJUaQnRfGDi4ZyqLSe3L0V7CuqZX9JHYXFdZ6vNRoMhFpMhIYYCbWYMZsMVNW20Np2\n8ixr6FjGdcyQeNIGRFJaZaeoooGiykaaWzvOay+6fARTRyf1SZ9yauGhZiYMG8CEYQOAjoVwdh+u\nJv/wcfIPV7Npd/kpnzcsLYapo5KYOiqJxNhTB6jb7ebTL0t585MDtDramTgskeuuHE20xXjW7RBd\nbje22maOVTRSVNlAUUUjxyobTxj2B0iKD+e8IQmcNzie0YPjdZmd+KXD5R0zzTMHBuZkNdAKa93i\n65WKWhxODpc1cKi0jkOl9TTY22hxtONoa6elrZ02ZzsJMWGkDYgkbUAkqQOiiI2ycKC4jj1Hqyks\nqsXhdHlezwAkJ0SQkRzFxGEDmDFu4Anv5+t++5q/9+tyu6motlPb6KCusbXjv02tHC1vYN+xWjr/\nT85MiWHM4HhiIkK+uswuBEuIiZzPD7PnSA3hoWauv3wE3xo3kKSkmHPqub7JwbHKRooqGthfXOdZ\ncQ86/n1lDIxmXGYC4zITGJYW6/PFMPz9Z+wNwdZzV/r99avbOFBSx+/vuahP55X0hFZY6wfCLGbP\njPXuGJ4Wy1XTM2hzujhQUoettpnUAZEMsgbuZI1g1HHuOZKUxJPnI9Q3Odi+38a2gkr2Hq3l8FfX\nsP5fE4YlcvNVo095CqUnYiItnsVsroaOdf/LGthzpJo9R2o4UFLH0fIG3t90lFCLiTEZ8YxMj2No\nagyDk6N1vbz0OZerY7Ja6oAJb1SMAAAckElEQVRIvw/uM9Fv7iASYjb2KPzF/8VEWrh4UhoXT0qj\nsbmN0qomGuwOGuxtHf9tbmNoagzTxyR79dpyk9HIsLRYhqXF8r2ZmTS3Otl3rJbdh6rJP1LtWVoX\nOj6MpFkjGZoawwXnJTMyPU7XvYvXlR1vwtHmYsjAwD3fDQpvkX4nKjzEbyYdhoeamTR8AJOGd5zD\nr6pr5lBpPYdK6zlcVs/R8gaOVTby6ZelpCdFcfmUQUw/L7lPL5GU4OJZnCWAz3eDwltE+tCA2HAG\nxIZz/phkoGNd+4MldXy8vYTt+2z8cUMBazce5MKJqVw6OU2z16XXfR3eOvIWEekRs8nIqIx4RmXE\nU13fwic7Svj0y1LWbz7K37YUMW1MEldOy2BwgP+iFf9xpLwe41dX9QQyhbeI+IWEmDB+cNEwvvet\nIWzeU8GHXxxj8+4KNu+uYFR6HFecn87E4QP8cjMfCQztLhfHKhpJHRAZ8KdmFN4i4lcsISYunJjK\n7Akp7D5czd++OMbuw9XsO1bLwIQIrjw/nW+NG0iIObB/+UrfK6uy43C6Anpxlk4KbxHxSwaDgXFf\nLQ1bbGvkb1uK2Ly7glc+2Mfb/zzM5VMGccnkNCLPssCMSKevF2cJ/PDWbgMi4vcGWaO45Tvn8egd\n3+Lq6Rm0Odt56x+H+Nnv/8Vb/zhEY3Obr0uUAHD0q8lqgwN8pjnoyFtEAkh8dCjZlwznu98awqdf\nlvLBliL+519H+GjrMS6fms4V09K1JKuc1pHyBkxGA+lJgb/xksJbRAJOeKiZq6ZncMnkND7dUcL6\n3K9D/NLJg7hsyqBeW0VO+gdnu4uiikbSrJH9Yr6EwltEAlZoiIkrzs/goqw0Nu4oYcPmoydcZnbF\ntPSAX4xDekdpVRPO9sBfWa2TwltEAl5oiIkrz8/gkqy0ky4zG5kex9zZmYzK0LLAway/rKzWSeEt\nIv3G/73M7MMvjpF/uJpHXt/B7Akp3JE9ydclio94wrsfXCYGCm8R6Ye+eZnZwdI6Xtmwj3/mlbHr\nUDULLx3O+WOStAlKkNl/rJYQs5G0AYG9slonXSomIv3asNRYfrF4KtkXD8Pe6mRVzm6eXLuT6voW\nX5cmfeR4XQslVU2MGRxPiLl/xF7/6EJE5AzMJiNXXzCYlT+7hLGZCeQfquahP2xhx36br0uTPrDr\n0HEAxg9N9HElvcer4Z2Tk8M111zDvHnz2Lhx4wmPvfbaayxcuJBFixbxq1/9yptliIgAMDAxknsW\nTOSmK0fhcLp45q+7eP1/C2lzunxdmnhR3sGvwnuYwvusampqWLlyJa+//jrPP/88H3/8seexxsZG\nXnrpJV577TXeeOMNDh48yJdffumtUkREPAwGAxdnpbHs5qmkDojko23F/GrNVsqr7b4uTbygzeli\nz9FqBiZEkBQX7utyeo3XwnvTpk3MmDGDqKgokpKSWLFiheexkJAQQkJCsNvtOJ1OmpubiY2N9VYp\nIiInGWSNYtnNU7lwYipFFY38vz9+wab8cl+XJb2s8FgtjjYXE/rRUTd4MbyLi4tpaWnh9ttv5/rr\nr2fTpk2ex0JDQ1myZAmXX345l1xyCRMnTiQzM9NbpYiInFJoiInFV4/m9u+PxWCAF/5nD394fy+t\njnZflya9xHO+u5+Ft1cvFautreXZZ5+ltLSUm266iU8++QSDwUBjYyOrVq3igw8+ICoqiptvvpmC\nggJGjx592teKj4/A7AdL2lmt/eMawa5Sv/1fsPV8qn6/Y41m8nkpPLrmCz7bVcbRykbuu2lqv9jA\nAoL7Z7z7SA1hFhMzswb1i2VRO3ktvBMTE8nKysJsNpORkUFkZCTV1dUkJiZy8OBB0tPTSUhIAGDq\n1Knk5+efMbxranx/PspqjcZma/B1GX1G/fZ/wdbzmfo1Az+7Lot1Gw/yv1uPcc+Tn3L9nJHMnpAS\n0NeEB/PPuLLGTomtkUnDB1DrBxnSE6f74OW1YfNZs2axefNmXC4XNTU12O124uM7lidMS0vj4MGD\ntLR0XGeZn5/PkCFDvFWKiEiXhJiNLLp8BHfNG4/ZZOTlDQU889dd1DU5fF2a9MCuQ9UA/e58N3jx\nyDs5OZkrr7ySBQsWALB06VLeeecdoqOjmTNnDrfccgs33XQTJpOJrKwspk6d6q1SRES6JWuklf+X\nHM0f1u/lywNVHHgxl5uuHMXU0Um+Lk26wXOJWD+6vruTwe12u31dRFf4w7BPMA8/BYNg6xeCr+fu\n9utyu/n7tmLWbjxIm9PFBecl829XjCQyLHD2DA/Wn7GjrZ27nvonSXHhrPj/pvu6rB473bC51jYX\nETkNo8HA5VPTGZuZwEvv72XzngoKi2u5a94EBveTrSX7q4KiWtqcrn43y7yTlkcVETmLlMRI/uuG\nyVw7K5Oa+lb++9VtbN6ta8L92a6vhswn9MMhc1B4i4h0iclo5JpZmdw9fwJmk4HV7+3hL38/gMsV\nEGceg4rb7SbvUBVhFhPDB/XPBcAU3iIi3TBx+ACW3jSVgQkRfLCliCfX7qSxuc3XZck3VNQ0Y6tt\nYWxmAmZT/4y5/tmViIgXpSRGsvSmqUwYlsjuw1/tUFaoHcr8Rd6BKqB/zjLvpPAWEemBiDAzd8+f\nwNzZmdQ3OXjmrV38/u1d1DW2+rq0oPev3eUYDQYm9tPJaqDZ5iIiPWY0GPjezEwmj0rilQ0FbN1n\nY8+RGhZcOpxZE1IwBvDKbIHqQHEtRRWNZI0YQGxUqK/L8RqFt4jIOUobEMn9N0xm444S1m08yMsb\nCli38SCjM+IYPTie0RnxpCRGBPQyq4Hiw9yjAFw4MdXHlXiXwltEpBcYDQYunTyIScMH8O5nh8k/\nXM3WfTa27us4Fx4baWHyKCsXnJfMsLRYHZV7QWtbO59uLyY+OpRxQxN8XY5XKbxFRHpRQkwY//7t\nMbjdbiprmyk4WkNBUS17jlTzyfYSPtleQmJMGNPPS2bySCvRESGEhpiwhBixmE0YjQr1ntpaUIm9\nxcmlkwdhMvbvKV0KbxERLzAYDCTHR5AcH8FFk9Jod7nYe7SG3N0VbCu0sX7zUdZvPnrS86IjQhg7\nJIFxQxMYm5lIbKTFB9UHpn/uLAVg9oQUH1fifQpvEZE+YDIaGZeZyLjMRG5sayfv4HH2FtXgcLTT\n6nThaGvH0dZORU0zm/dUsHlPBQCDk6MZlRFHelIUg6xRpA6I6Ff7UveWsuNNFBbXMWmEFWtcuK/L\n8TqFt4hIH7OEmJg6OumUu5S53W5KqprIP1TNrkPH2V9cy9GKrzcWMRoMJCeEk5EczeDkaAYnR5Ex\nMDqgNkvxhn/mlQFwxfTBPq6kbyi8RUT8iMFgYJC14yj7qukZtDraOVbZyDFbI8WVjRTbOv6UHbeT\n+9XROcCA2DBmT0jhyvMzsIQE15G5s93F57vKiAwzc8H4gdTW2H1dktcpvEVE/FjoV+tzf3ONbrfb\nja2uhaLyBo5WNHC0vIGDpfW8/c/DfLqzlB9cNIzp5yUHzYz2L/dX0WBvY87U9KA5paDwFhEJMAaD\ngaS4cJLiwj1D782tTt7fdJQPvzjGC+/t4aOtxSy6bMRp94PuT/6R1zFR7cKJ/X+iWqf+PZdeRCRI\nhIeamX/xMH5163SmjU7icFk9//3qNtZs2Ivb3X93Pjte18LuQ9UMS4shzRrl63L6jMJbRKQfscaF\nc8e143jghilY48L4y0eFvPg/e3C2u3xdmld8kFuEG7hwQv9eUe3/UniLiPRDwwfF8uBNUxk1OJ5N\nuyt44s0vaWrpX1uXVlTb2fhlCUnx4cwYN9DX5fQphbeISD8VE2HhV3fMZPJIKwVFtfz3mm1U1TX7\nuqxes+7Tg7S73My/aFi/3bf7dIKrWxGRIBMaYuLOa8cxZ2o6Zcft/OpP2yj6xnXjgepASR3b9tkY\nlhrDlFFWX5fT5xTeIiL9nNFoYNHlI1h02Qjqmhw8+voODhTX+bqsHnO73fzl7wcAWHDp8KDcrU3h\nLSISJOZMS+fW751Hi6Odx97cQf6h474uqUe2F1ZxoKSOySOtjBgU5+tyfELhLSISRGaMHch/zhuP\n2w1Prctja0Glr0vqFme7i3WfHsRoMPCDi4b6uhyfUXiLiASZSSMGcM+CiYSYjTz3bj7/+Go3rkDw\nj52lVFTbuSgrlZTESF+X4zMKbxGRIDQqI56fLcoiMiyElzcU8Je/H8Dl8u/FXJpa2nj3s8OEWkxc\nMzPT1+X4lMJbRCRIZabE8F83TGZgQgQfbCniybU7aWz2z2vB25wuVr61iwZ7G9+dMTjo9zn3anjn\n5ORwzTXXMG/ePDZu3HjCY2VlZSxatIj58+fzi1/8wptliIjIaaQkRrL0pqlMGJbI7sPVPPzKVopt\njb4u6wRut5uXN+yloKiWKaOsXH1BcGz7eSZeC++amhpWrlzJ66+/zvPPP8/HH398wuO/+c1v+OEP\nf8i6deswmUyUlgbOORcRkf4kIszM3T+YwHe/NZjK2mZ+9adtbN5T7jdror/72WE27a5gWGoMt373\nvKDZLe1MvBbemzZtYsaMGURFRZGUlMSKFSs8j7lcLrZt28all14KwEMPPURqanCtSysi4k+MRgPz\nLhzGHdeOw42b1Tl7ePhP29hzpNqndX2WV0bO50ewxoVx1/wJQbdX+el4LbyLi4tpaWnh9ttv5/rr\nr2fTpk2ex6qrq4mMjOTXv/41ixYt4vHHH/dWGSIi0g3TRiex/N/PZ+ooK4fL6nnsz1/y2zd2cLis\nvs9r2XOkmlc+KCAyzMxPFkwiJiK4z3N/k8HtpXGR1atXs337dp599llKS0u56aab+OSTTzAYDNhs\nNubMmUNOTg5paWncdttt3HjjjVx88cWnfT2nsx1zkGyyLiLiD/Yfq2HN+r3sKLQBcMG4gVx/5Wgy\nU2O9+r5ut5sPc4t48d1dONvdPHz7txg7NNGr7xlozN564cTERLKysjCbzWRkZBAZGUl1dTWJiYnE\nx8eTmppKRkYGADNmzGD//v1nDO+aGru3Su0yqzUamy3w1wTuKvXb/wVbz8HWL5xbz3FhZu6aN56C\nozX89R8H2Zxfzub8cqaMtPL9WZkMSur9/bPrGlv544YC8g4eJzzUzB3fP4+kaEuXe+hvP2OrNfqU\n93tt2HzWrFls3rwZl8tFTU0Ndrud+Ph4AMxmM+np6Rw5cgSA3bt3k5kZ3NfsiYj4q9GD43nghin8\nZMFEMlNi2FZo4xd/2MLv38ln75HqXtsrfGtBJcte2kLeweOMGRzPilvOJ2tk8G060hVeO/JOTk7m\nyiuvZMGCBQAsXbqUd955h+joaObMmcMDDzzA/fffj9vtZuTIkZ7JayIi4n8MBgPjhyYyLjOBvIPH\neeezw2wtqGRrQSURoWYmDE8ka4SVcZkJhId2PVrcbjcHS+v5ILeI7YU2LGYj/zZnJJdMTtOs8jPw\n2jnv3uYPwyD9bTjmbNRv/xdsPQdbv+C9nt1uNwVHa9hWaOPLA1VU17cCYDIaGDEolrGZCYzLTCQ9\nOeqUIexsd/FFQSUfbT3G4bKO+oalxvDD74w5p2VP+9vP+HTD5l478hYRkf7LYDAwZkgCY4Yk8G9z\nRlJU0ciO/TZ2HjhOQVEtBUW1/PXTQ0SFhzA8LRaTyYDb3RH6bjccLqunrsmBAcgaMYDLp6YzOiMu\nKLf37AmFt4iInBODwcDggdEMHhjNtbOHUm93sPdIDbsPV7P7SDVfHqg66TnhoSaumJbOpVMGkRQX\n7oOqA5vCW0REelVMhIXp5yUz/bxk3G63Z710g8GA0QBgwBJixGzS9ho9pfAWERGvMRgMRGtxlV6n\njz0iIiIBRuEtIiISYBTeIiIiAUbhLSIiEmAU3iIiIgFG4S0iIhJgFN4iIiIBRuEtIiISYBTeIiIi\nAUbhLSIiEmAU3iIiIgEmYPbzFhERkQ468hYREQkwCm8REZEAo/AWEREJMApvERGRAKPwFhERCTAK\nbxERkQAT1OFdWFjI5Zdfzquvvtrl55SVlXHjjTdy/fXX86Mf/QiHwwFAQUEB8+bNY968eaxcudJb\nJZ+z3ux57Nix3HjjjZ4/7e3t3iq7x3qz30733HMP999/f2+X2mt6s+dnn32WhQsXsmDBAn7/+997\nq+Rz0pv9rl+/nvnz57NgwQKefPJJb5V8Tnqz37q6Om655Rbuvvtub5V7TnrS6ze9+OKLzJ8/n+zs\nbD799FMAGhoauO2221i0aBG33HILtbW1vVlynwna8Lbb7axYsYIZM2Z063lPP/00119/Pa+//jqD\nBw9m3bp1ACxbtowVK1awbt06Dh48SHNzszfKPie93XNUVBRr1qzx/DGZTN4ou8d6u1+Azz//nKKi\not4utdf0Zs/FxcUUFhby5ptv8sYbb/DOO+9QUVHhpcp7pjf7bW5u5rHHHuPll1/mzTff5F//+hcH\nDhzwUuU909v/ph966CGmTJnijVLPWU977XTs2DHWr1/P66+/zqpVq/j1r39Ne3s7r7zyCueffz5v\nvPEGV1xxBS+88EIvV943gja8LRYLL7zwAklJSZ77Dhw4wE033cTNN9/MnXfeSX19/UnPy83N5bLL\nLgPgkksuYdOmTVRVVWG32xk7dixGo5EnnniC8PDwPuulq3qz50DQ2/06HA6ee+457rjjjr5poAd6\ns+dBgwbx9NNPAx1HaAaDgaioqL5ppIt6s9/w8HBycnKIiorCYDAQFxfnd0dlvf1v+uGHH/bb8O5u\nr7m5uTzzzDMn3J49ezYWi4WEhATS0tI4cOAAmzZtYs6cOUBg/T77v4I2vM1mM2FhYSfct2LFCn75\ny1/yyiuvMHPmTF577bWTntfc3IzFYgEgMTERm81GSUkJsbGx3H///Vx33XW8/PLLfdFCt/Vmz9AR\nZvfeey/XXXcdf/zjH73fQDf1dr+rVq1i0aJFfhdg39TbPUPHL/jvfve73HnnnURGRnq3gW7q7X47\nf7b79u2jpKSEiRMnermD7vFWv/6op712qqqqIiEhwXM7ISEBm812wv2JiYlUVlZ6pwEvM/u6AH+S\nl5fHsmXLgI5gGj9+/Bm/vnNlWbfbTXFxMStXriQsLIyFCxcyc+ZMRowY4fWaz1VPewb4+c9/zjXX\nXIPBYOCGG25g6tSpZ32+r/W03yNHjpCfn89dd91Fbm6u1+vsTefyMwZYunQpd911FzfeeCOTJ08m\nPT3da7X2hnPt98iRI/z0pz/l8ccfJyQkxGt19pZz7TeQnKrXrVu38tRTT1FfX09DQwNbtmzh8ssv\nP+m5p+o7kL8XCu9vCA8P509/+hMGg8Fz344dO3jiiScAeOyxx4iIiKClpYWwsDAqKipISkoiMTGR\nESNGEB8fD8CUKVPYv39/QIR3T3sGWLRokec5F1xwAYWFhX4f3j3td+PGjZSWlrJgwQIaGxuprq7m\nhRde4NZbb/VVK13W057Lysqoqqpi/PjxxMbGMnnyZHbt2uX34X0u/6bLy8tZsmQJjz76KGPGjPFJ\n/d11Lv0GmlP1CrBmzRpyc3PZsmULd911FwBvvfUWhw8f9nxNZ99JSUnYbDaio6MD+nsRtMPmpzJ6\n9Gj+8Y9/APD++++zadMmsrKyPBOykpOT+da3vsXf/vY3AD788ENmz55Neno6TU1N1NbW4nK52Lt3\nL0OHDvVlK13W054PHTrEvffei9vtxul0sn379oD4sNLTfhcvXsx7773HX/7yFx566CEuvvjigAhu\n6HnP1dXVLF++HKfTSXt7O7t37yYzM9OXrXRJT/sFePDBB1m+fDljx471Wf3ddS79BppT9Xo6F1xw\nARs3bsThcFBRUUFlZSXDhw9n5syZfPDBB0Bgfy+Cdlex/Px8HnnkEUpKSjCbzSQnJ/PjH/+Yxx9/\nHKPRSGhoKI8//jhxcXEnPK+yspL77ruP1tZWUlNT+fWvf01ISAg7d+7k4YcfxmAwMHv2bM+nP3/S\n2z3/9re/ZfPmzRiNRi699FK/m8jV2/12ys3N5e233+Y3v/lNX7d0Vr3d86pVq/joo49wu91cfPHF\n/Od//qePOju13uy3uLiYa6+9lgkTJni+bvHixZ6JXv6gN/s1Go0sXryY+vp6KioqGDFiBHfeeWeP\nZ3f3tp72+k1r1qzhvffew2Aw8OMf/5gZM2bQ1NTEz372M2pra4mJieG3v/0t0dHRfdhZ7wja8BYR\nEQlUGjYXEREJMApvERGRAKPwFhERCTAKbxERkQCj8BYREQkwCm8RP1VcXMyFF17Yp+/ZW7vDjRo1\nihtuuMGz49yCBQv48MMPz/q89957D5fLdc7vL9LfaYU1EfFYs2ZNr73Wyy+/jNnc8SumqqqK73//\n+5x//vlnvC73mWee4eqrr8Zo1HGFyJkovEUC0Pr163n11Vdxu90kJCTw8MMPEx8fz+uvv867775L\nSEgIoaGhPPnkk8TExHDppZdy9dVXc+zYMX7+859zxx13MGvWLPLy8mhqamLVqlUkJyczatQodu/e\nzXPPPUdtbS3l5eUcPXqU6dOns2zZMlpbW7nvvvsoKSlh4MCBmEwmZs6cSXZ29hnrHTBgAFarlaKi\nImJiYnjooYc4dOgQDoeDiRMnsnTpUp5++mmOHj3K4sWLefbZZykoKGDlypW43W7MZjMrVqzw+6VZ\nRfqKPt6KBJiysjKef/55Xn75Zd544w3OP/98Vq1aBUBraysvvfQSr776KmlpaeTk5HieN2TIEM8W\nnwcPHmTevHm89tprjBkzhg0bNpz0Pnv27OHpp59m3bp1vPXWW9TV1ZGTk4PT6WTt2rX84he/4PPP\nP+9Szfn5+VRWVjJs2DDq6uoYNWoUr732GmvXruWzzz6jsLCQu+++G+g4Yg8NDeWhhx7imWee4dVX\nX+WGG27g0UcfPddvnUi/oSNvkQCzY8cObDYbt9xyC9Cxu9KgQYMAiIuL47bbbsNoNFJSUoLVavU8\nLysry/P3+Ph4z1r0qampp9y3esqUKZhMJkwmE/Hx8dTV1bF3717OP/98AKxW6xn3gl68eDEGg4Gq\nqirCwsJ4/vnniYyMJCwsjLKyMhYuXIjFYsFms1FTU3PCc/fv34/NZvMsM9ze3n7SZhQiwUzhLRJg\nLBYLEyZM8BxtdyovL+eRRx7h/fffJzExkUceeeSEx7+5PrvJZDrhsVOtknyqr3G5XCecjz7TuenO\nc955eXncd999jBw5EujYUGLXrl289tprmM1m5s2bd8oeU1NTe/UcvEh/omFzkQAzfvx48vLysNls\nAGzYsIGPPvqI48ePEx8fT2JiIrW1tXz22Wc4HI5efe+hQ4eyY8cOAI4fP862bdvO+pwJEyYwa9Ys\nfve733mel5mZidlsJj8/n6KiIk+dBoMBp9PJkCFDqKmpobCwEIAvvviCN998s1d7EQlkOvIW8WPV\n1dXceOONntvjx4/n5z//OQ8++CD/8R//QXh4OGFhYTzyyCMkJCQwePBg5s+fT0ZGBnfffTfLly/n\noosu6rV65s2bx8aNG1m4cCGDBg1i6tSpJx2hn8qPf/xjrrnmGq688kquuuoqbr/9dm644QYmT57M\nD3/4Qx5++GH+8pe/MHv2bH7wgx/w3HPP8dvf/pYHH3yQ0NBQAH75y1/2Wh8igU67iolIl1VUVLB9\n+3auvvpqXC4Xc+fOZfny5SecTxcR79ORt4h0WXR0NOvXr+ell17CYDBw4YUXKrhFfEBH3iIiIgFG\nE9ZEREQCjMJbREQkwCi8RUREAozCW0REJMAovEVERAKMwltERCTA/P+Y3Pz5yyFdNQAAAABJRU5E\nrkJggg==\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f5ed99a6a58>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "x7fO0cuVET8W",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "learn.save_encoder('ft_enc')\n",
        "#Finally we save the encoder to be able to use it for classification in the next section."
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "RTEY6pBu6Koh",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "##Building a classifier\n",
        "We now use the data_clas object we created earlier to build a classifier with our fine-tuned encoder. The learner object can be done in a single line.\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "uSkXI7HJET_r",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "learn = text_classifier_learner(data_clas, drop_mult=0.7)\n",
        "learn.load_encoder('ft_enc')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "1VaunwDAMmLV",
        "colab_type": "code",
        "outputId": "05d2db52-38a0-4076-f7af-951740c26481",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 493
        }
      },
      "cell_type": "code",
      "source": [
        "learn.lr_find()\n",
        "learn.recorder.plot()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='2' class='' max='3', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      66.67% [2/3 00:41<00:20]\n",
              "    </div>\n",
              "    \n",
              "<table style='width:300px; margin-bottom:10px'>\n",
              "  <tr>\n",
              "    <th>epoch</th>\n",
              "    <th>train_loss</th>\n",
              "    <th>valid_loss</th>\n",
              "    <th>accuracy</th>\n",
              "  </tr>\n",
              "  <tr>\n",
              "    <th>1</th>\n",
              "    <th>0.718614</th>\n",
              "  </tr>\n",
              "  <tr>\n",
              "    <th>2</th>\n",
              "    <th>0.745654</th>\n",
              "  </tr>\n",
              "</table>\n",
              "\n",
              "\n",
              "    <div>\n",
              "        <style>\n",
              "            /* Turns off some styling */\n",
              "            progress {\n",
              "                /* gets rid of default border in Firefox and Opera. */\n",
              "                border: none;\n",
              "                /* Needs to be in here for Safari polyfill so background images work as expected. */\n",
              "                background-size: auto;\n",
              "            }\n",
              "            .progress-bar-interrupted, .progress-bar-interrupted::-webkit-progress-bar {\n",
              "                background: #F44336;\n",
              "            }\n",
              "        </style>\n",
              "      <progress value='0' class='progress-bar-interrupted' max='44', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
              "      Interrupted\n",
              "    </div>\n",
              "    "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "LR Finder is complete, type {learner_name}.recorder.plot() to see the graph.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfsAAAFYCAYAAABUA1WSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f5ed9922d30>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "s7GTTw1GMl2F",
        "colab_type": "code",
        "outputId": "cab75181-040e-4099-f6bb-7fe8838379d9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 94
        }
      },
      "cell_type": "code",
      "source": [
        "learn.fit_one_cycle(1, 1e-2)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "Total time: 00:32 <p><table style='width:300px; margin-bottom:10px'>\n",
              "  <tr>\n",
              "    <th>epoch</th>\n",
              "    <th>train_loss</th>\n",
              "    <th>valid_loss</th>\n",
              "    <th>accuracy</th>\n",
              "  </tr>\n",
              "  <tr>\n",
              "    <th>1</th>\n",
              "    <th>0.526815</th>\n",
              "    <th>0.417662</th>\n",
              "    <th>0.835443</th>\n",
              "  </tr>\n",
              "</table>\n"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "XZHQ9N0TK0NA",
        "colab_type": "code",
        "outputId": "10ccdf89-ae28-4759-b997-fc42270198cf",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 142
        }
      },
      "cell_type": "code",
      "source": [
        "# get predictions\n",
        "preds, targets = learn.get_preds()\n",
        "\n",
        "predictions = np.argmax(preds, axis = 1)\n",
        "pd.crosstab(predictions, targets)"
      ],
      "execution_count": 0,
      "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>col_0</th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>row_0</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>172</td>\n",
              "      <td>16</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>62</td>\n",
              "      <td>224</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "col_0    0    1\n",
              "row_0          \n",
              "0      172   16\n",
              "1       62  224"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    },
    {
      "metadata": {
        "id": "2ovZaadqKv3x",
        "colab_type": "code",
        "outputId": "a00ea6d8-b237-4b9e-803d-5b6a178bd84c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "print(\"Process completed \")"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Process completed \n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}