Rafid_Abyaad_ULMFiT_test.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/kaggle/input/quora-insincere-questions-classification/test.csv\n",
      "/kaggle/input/quora-insincere-questions-classification/train.csv\n",
      "/kaggle/input/quora-insincere-questions-classification/sample_submission.csv\n",
      "/kaggle/input/quora-insincere-questions-classification/embeddings/wiki-news-300d-1M/wiki-news-300d-1M.vec\n",
      "/kaggle/input/quora-insincere-questions-classification/embeddings/glove.840B.300d/glove.840B.300d.txt\n",
      "/kaggle/input/quora-insincere-questions-classification/embeddings/paragram_300_sl999/paragram_300_sl999.txt\n",
      "/kaggle/input/quora-insincere-questions-classification/embeddings/paragram_300_sl999/README.txt\n",
      "/kaggle/input/quora-insincere-questions-classification/embeddings/GoogleNews-vectors-negative300/GoogleNews-vectors-negative300.bin\n",
      "/kaggle/input/quoratest/custom.css\n",
      "/kaggle/input/quoratest/__results__.html\n",
      "/kaggle/input/quoratest/__output__.json\n",
      "/kaggle/input/quoratest/__notebook__.ipynb\n",
      "/kaggle/input/quoratest/gen2.pth\n",
      "/kaggle/input/quoratest/__results___files/__results___8_1.png\n",
      "/kaggle/input/quoratest/models/tmp.pth\n"
     ]
    }
   ],
   "source": [
    "# This Python 3 environment comes with many helpful analytics libraries installed\n",
    "# It is defined by the kaggle/python docker image: https://github.com/kaggle/docker-python\n",
    "# For example, here's several helpful packages to load in \n",
    "\n",
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
    "\n",
    "# Input data files are available in the \"../input/\" directory.\n",
    "# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n",
    "\n",
    "import os\n",
    "for dirname, _, filenames in os.walk('/kaggle/input'):\n",
    "    for filename in filenames:\n",
    "        print(os.path.join(dirname, filename))\n",
    "\n",
    "# Any results you write to the current directory are saved as output."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "_cell_guid": "79c7e3d0-c299-4dcb-8224-4455121ee9b0",
    "_uuid": "d629ff2d2480ee46fbb7e2d37f6b5fab8052498a"
   },
   "outputs": [],
   "source": [
    "from fastai import *\n",
    "from fastai.text import *\n",
    "from fastai.callbacks import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_lm = TextLMDataBunch.from_csv('/kaggle/input/quora-insincere-questions-classification/','train.csv',text_cols= 'question_text')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>idx</th>\n",
       "      <th>text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>xxbos xxmaj who are some unknown teenage xxmaj quorans ? xxbos xxmaj how dimensionality is reduced based on autoencoder ? xxbos xxmaj was xxmaj france a part of the alliance with xxmaj britain , the xxup usa and the xxup ussr in 1945 after xxup ww2 ? xxbos xxmaj what question can i discuss as a topic on ' xxunk ' for a literature review ? xxbos xxmaj what all</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>martial artist ever ? xxbos xxmaj what are the 5 things that would make cold calling effective if one is seeking a meeting to speak about investment options or wealth management ? xxbos xxmaj does cubical xxunk depend on the initial volume ? xxbos xxmaj how can i install three xxup os on my laptop ? xxbos xxmaj what are some factors that affect libido ? xxbos xxmaj if the</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>long - xxmaj term xxmaj fat xxmaj loss ? xxbos xxmaj is it possible to become an xxup ies officer without coaching ? xxbos xxmaj what are the xxmaj paytm share price fluctuations over the past 2 years and why ? xxbos xxmaj how many squats must i do to burn 100 calories ? xxbos xxmaj could it be assumed that xxmaj polonius had some culpability in the murder of</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>had not broken down , life would not have developed on xxmaj earth . xxmaj why ? xxbos xxmaj at what stage in the drug development pipeline is a new lead compound typically patented ? xxbos xxmaj when will xxmaj sm - xxup j700f xxmaj android version xxmaj xxunk xxunk ? xxbos xxmaj what is the best way to have xxmaj many xxunk on xxmaj facebook group ? xxbos xxmaj</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>xxmaj what should i include in my appeal ? xxbos xxmaj is it alright to address a steward / flight attendant by their name ? xxmaj this question is based on varied comments shot for an answer by xxmaj jake xxmaj williams . xxbos xxmaj what is the best kind of xxunk and meditation for mental disorders ? xxbos xxmaj how can the government tolerate xxmaj violence created by xxmaj</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_lm.show_batch()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "learner= language_model_learner(data_lm,AWD_LSTM,metrics=accuracy,path='/kaggle/working')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "learner.load('/kaggle/input/quoratest/gen2');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "learner.save_encoder('fine_tuned_enc')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_class = TextClasDataBunch.from_csv('/kaggle/input/quora-insincere-questions-classification/','train.csv',\n",
    "                                        valid_pct=0.05, vocab=data_lm.vocab,label_cols='target')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>text</th>\n",
       "      <th>target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>xxbos xxmaj what is [ math ] \\ frac { \\ xxunk } } \\ tan ( \\ tan ( { \\ boxed { \\ xxunk } \\ sum _ { \\ xxunk { \\ infty } \\ xxunk } \\ sum _ { \\ xxunk { \\ infty } \\ underbrace { \\ xxunk } } _ { \\ xxunk 's xxmaj law of xxmaj theoretical xxmaj probability .</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>xxbos xxmaj there 's this girl that i 've known since i was 5 ( 20 now ) . i 've had a crush on her since i was 10 . xxmaj what would be the best way to go about asking her out ? xxmaj at this point , i really just want to know whether or not i would \" have a shot \" with her .</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>xxbos xxmaj what is the etymology of the xxmaj spanish word \" mar \" , the xxmaj english word \" ocean \" or its xxmaj spanish cognate \" xxunk \" and the xxmaj english word \" sea \" ? xxmaj are all related to the concept of a world river or the like , the same way the word \" ocean \" is ?</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>xxbos xxmaj if one have n’t being getting enough sleep for 3 - 4 mnths ( constantly sleeping from 2 - 7 am ) , and now wants to make up for the bad habit , by getting enough and proper rest ( that can remedy any damage that might have been done ) , how does he go about it ?</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>xxbos xxmaj you have been locked in a room by your kidnapper for a long time . xxmaj one day , an opportunity for you to escape appears and as you escape , you stumble across two things : a cellphone and a gun . xxmaj if you could only take one with you , which would you take ?</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "data_class.show_batch()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifier = text_classifier_learner(data_class, AWD_LSTM,drop_mult = 0.1,path='/kaggle/working')\n",
    "classifier.metrics = [accuracy,FBeta(beta=1)]\n",
    "classifier.load_encoder('fine_tuned_enc');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "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='8' class='' max='10', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      80.00% [8/10 49:25<12:21]\n",
       "    </div>\n",
       "    \n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>f_beta</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>0.225814</td>\n",
       "      <td>0.447693</td>\n",
       "      <td>0.929456</td>\n",
       "      <td>0.404550</td>\n",
       "      <td>06:26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.214352</td>\n",
       "      <td>0.196970</td>\n",
       "      <td>0.937311</td>\n",
       "      <td>0.025238</td>\n",
       "      <td>06:04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.195434</td>\n",
       "      <td>0.218698</td>\n",
       "      <td>0.939425</td>\n",
       "      <td>0.278366</td>\n",
       "      <td>06:09</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.218628</td>\n",
       "      <td>0.199128</td>\n",
       "      <td>0.939516</td>\n",
       "      <td>0.315425</td>\n",
       "      <td>06:26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.199436</td>\n",
       "      <td>0.245087</td>\n",
       "      <td>0.937756</td>\n",
       "      <td>0.380996</td>\n",
       "      <td>05:58</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.219488</td>\n",
       "      <td>0.740582</td>\n",
       "      <td>0.939853</td>\n",
       "      <td>0.178586</td>\n",
       "      <td>05:54</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.220863</td>\n",
       "      <td>0.237752</td>\n",
       "      <td>0.937373</td>\n",
       "      <td>0.058905</td>\n",
       "      <td>06:26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.210498</td>\n",
       "      <td>0.234848</td>\n",
       "      <td>0.939501</td>\n",
       "      <td>0.312511</td>\n",
       "      <td>05:58</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table><p>\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='19271' class='' max='19387', style='width:300px; height:20px; vertical-align: middle;'></progress>\n",
       "      99.40% [19271/19387 05:46<00:02 0.2013]\n",
       "    </div>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "classifier.fit(10,1e-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "classifier.recorder.plot_metrics()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifier.save('/kaggle/working/classifier')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}