{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Proofbird grammar model (French): quick start\n",
    "\n",
    "Run [`proofbird/gector-bert-fr`](https://huggingface.co/proofbird/gector-bert-fr) in a few lines. It is small (about 450 MB), runs on a CPU in about\n",
    "30 ms a sentence, and needs no API key."
   ]
  },
  {
   "cell_type": "code",
   "metadata": {},
   "execution_count": null,
   "outputs": [],
   "source": [
    "!pip install -q onnxruntime tokenizers huggingface_hub"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Load the model and correct a sentence\n",
    "\n",
    "`quickstart.py`, published with the model, downloads it, runs it, and returns the corrected text plus one suggestion per\n",
    "change (character position, original, replacement)."
   ]
  },
  {
   "cell_type": "code",
   "metadata": {},
   "execution_count": null,
   "outputs": [],
   "source": [
    "import importlib.util\n",
    "from huggingface_hub import hf_hub_download\n",
    "\n",
    "path = hf_hub_download(\"proofbird/gector-bert-fr\", \"quickstart.py\")\n",
    "spec = importlib.util.spec_from_file_location(\"quickstart\", path)\n",
    "quickstart = importlib.util.module_from_spec(spec)\n",
    "spec.loader.exec_module(quickstart)\n",
    "\n",
    "model = quickstart.Gector(\"proofbird/gector-bert-fr\")\n",
    "corrected, suggestions = model.correct(\"Ma sœur sont très content.\")\n",
    "print(corrected)\n",
    "suggestions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. See every pass\n",
    "\n",
    "The model does not write text. For every word it picks one of 5 000 edits; the corrected sentence goes back in until\n",
    "nothing changes, so connected errors are fixed one after the other. `verbose=True` prints what each pass sent and got\n",
    "back."
   ]
  },
  {
   "cell_type": "code",
   "metadata": {},
   "execution_count": null,
   "outputs": [],
   "source": [
    "corrected, suggestions = model.correct(\"Ma sœur sont très content.\", verbose=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Your own text\n",
    "\n",
    "`model.passes` holds the same per-pass detail as a list of dictionaries, if you want to build on it."
   ]
  },
  {
   "cell_type": "code",
   "metadata": {},
   "execution_count": null,
   "outputs": [],
   "source": [
    "text = \"Les enfants joue dans le jardin.\"\n",
    "print(model.correct(text)[0])\n",
    "model.passes[0][\"words\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "Published by Patrick Andrew Hartnett. The [Proofbird](https://proofbird.app) writing assistant for Windows uses this\n",
    "model on-device and adds the guards in `gector-guards.json` and an explanation for each change. If you build on this\n",
    "model, a link back is appreciated."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "name": "python"
  },
  "colab": {
   "provenance": []
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}