Hatespeech classifer based on fastai lesson 3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Hatespeech Classifier**\n",
    "\n",
    "Data from https://github.com/aitor-garcia-p/hate-speech-dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai.text import *"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Obviously, modify the path below to reflect where your data is stored"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "path = pathlib.PosixPath(Path.cwd())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "bs=48\n",
    "hate_data_lm = (TextList.from_folder(path)\n",
    "           #Inputs: all the text files in path\n",
    "            .filter_by_folder(include=['unsup','train','test']) \n",
    "           #We may have other temp folders that contain text files so we only keep what's in train and test\n",
    "            .split_by_rand_pct(0.1)\n",
    "           #We randomly split and keep 10% (10,000 reviews) for validation\n",
    "            .label_for_lm()           \n",
    "           #We want to do a language model so we label accordingly\n",
    "            .databunch(bs=bs))\n",
    "hate_data_lm.save('hate_data_lm.pkl')\n",
    "hate_data_lm = load_data(path, 'hate_data_lm.pkl', bs=bs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "hate_data_lm.show_batch()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn = language_model_learner(hate_data_lm, AWD_LSTM, drop_mult=0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.lr_find()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.recorder.plot(skip_end=15, suggestion=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.fit_one_cycle(1, 7e-2, moms=(0.8,0.7))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save('hate_fit_head')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.load('hate_fit_head')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.unfreeze()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.fit_one_cycle(10, 1e-3, moms=(0.8,0.7)) # gets about 0.28"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save('hate-fine-tuned')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save_encoder('hate-fine-tuned_enc')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('annotations_metadata.csv')\n",
    "df.head()\n",
    "df2 = df.set_index('file_id')\n",
    "df2.head()\n",
    "df2.loc['12834217_1','label']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "hate_data_clas = (TextList.from_folder(path, vocab=hate_data_lm.vocab)\n",
    "             #grab all the text files in path\n",
    "             .split_by_folder(valid='train')\n",
    "             #split by train and valid folder (that only keeps 'train' and 'test' so no need to filter)\n",
    "             #.label_from_csv('/home/andy/Documents/nlp/data/hate-speech-dataset/annotations_metadata.csv')\n",
    "             #.label_from_func(lambda x : df2['label'].iloc[ (os.path.basename(x))[:-4] ])\n",
    "             .label_from_func(lambda x : df2.loc[(os.path.basename(x))[:-4], 'label'])\n",
    "             .databunch(bs=bs))\n",
    "\n",
    "hate_data_clas.save('hate_data_clas.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "hate_data_clas.show_batch()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn = text_classifier_learner(hate_data_clas, AWD_LSTM, drop_mult=0.5)\n",
    "learn.load_encoder('hate-fine-tuned_enc')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.lr_find()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.recorder.plot(suggestion=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.fit_one_cycle(1, 3.63e-3, moms=(0.8,0.7)) # from suggestion, gets about 0.76"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save('hate-first')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.load('hate-first')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.freeze_to(-3)\n",
    "learn.fit_one_cycle(1, slice(5e-3/(2.6**4),5e-3), moms=(0.8,0.7)) # gets about 0.80 accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save('hate-unfrozen-to-3')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.lr_find()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.recorder.plot(suggestion=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.unfreeze()\n",
    "learn.fit_one_cycle(2, 2e-5, moms=(0.8,0.7))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.predict(\"Insert a sample sentence to be classified here.\")"
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}