{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Masked LM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-info\">\n",
    "\n",
    "This tutorial is available as an IPython notebook at [malaya-speech/example/mlm](https://github.com/huseinzol05/malaya-speech/tree/master/example/mlm).\n",
    "    \n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-warning\">\n",
    "\n",
    "This module is not language independent, so it not save to use on different languages. Pretrained models trained on hyperlocal languages.\n",
    "    \n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Purpose"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When doing CTC or RNNT beam decoding, we want to add language bias during find the optimum alignment using Mask Model."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### List available Masked models\n",
    "\n",
    "We provided a few Masked models for our ASR models,"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import malaya_speech"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "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>Size (MB)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>mesolitica/bert-base-standard-bahasa-cased</th>\n",
       "      <td>443</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mesolitica/roberta-base-standard-bahasa-cased</th>\n",
       "      <td>310</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                               Size (MB)\n",
       "mesolitica/bert-base-standard-bahasa-cased           443\n",
       "mesolitica/roberta-base-standard-bahasa-cased        310"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "malaya_speech.language_model.available_mlm()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load Masked Model\n",
    "\n",
    "```python\n",
    "def mlm(model: str = 'mesolitica/bert-base-standard-bahasa-cased', force_check: bool = True, **kwargs):\n",
    "    \"\"\"\n",
    "    Load Masked language model.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    model: str, optional (default='mesolitica/bert-base-standard-bahasa-cased')\n",
    "        Check available models at `malaya_speech.language_model.available_mlm()`.\n",
    "    force_check: bool, optional (default=True)\n",
    "        Force check model one of malaya model.\n",
    "        Set to False if you have your own huggingface model.\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    result: malaya_speech.torch_model.mask_lm.LM class\n",
    "    \"\"\"\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "lm = malaya_speech.language_model.mlm()\n",
    "lm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "malaya-speech Masked LM need to combine with `pyctcdecode` to decode CTC logits."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Use pyctcdecode\n",
    "\n",
    "#### From PYPI\n",
    "\n",
    "```bash\n",
    "pip3 install pyctcdecode==0.1.0 pypi-kenlm==0.1.20210121\n",
    "```\n",
    "\n",
    "#### From source\n",
    "\n",
    "Check https://github.com/kensho-technologies/pyctcdecode how to build from source incase there is no available wheel for your operating system.\n",
    "\n",
    "Building from source should only take a few minutes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pyctcdecode import Alphabet, BeamSearchDecoderCTC\n",
    "from malaya_speech.utils.char import CTC_VOCAB\n",
    "\n",
    "labels = CTC_VOCAB + ['_']\n",
    "ctc_token_idx = len(CTC_VOCAB)\n",
    "alphabet = Alphabet.build_alphabet(labels, ctc_token_idx=ctc_token_idx)\n",
    "decoder = BeamSearchDecoderCTC(alphabet, lm)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.10"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}