test/w2ner_bert.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "import os\n",
    "import jieba\n",
    "import torch\n",
    "import pickle\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import pandas as pd\n",
    "\n",
    "from ark_nlp.model.ner.w2ner_bert import W2NERBert\n",
    "from ark_nlp.model.ner.w2ner_bert import W2NERBertConfig\n",
    "from ark_nlp.model.ner.w2ner_bert import Dataset\n",
    "from ark_nlp.model.ner.w2ner_bert import Task\n",
    "from ark_nlp.model.ner.w2ner_bert import get_default_model_optimizer\n",
    "from ark_nlp.factory.lr_scheduler import get_default_linear_schedule_with_warmup\n",
    "from ark_nlp.model.ner.w2ner_bert import Tokenizer\n",
    "from ark_nlp.factory.utils.seed import set_seed\n",
    "set_seed(42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 目录地址\n",
    "train_data_path = '../data/source_datasets/CMeEE/CMeEE_train.json'\n",
    "dev_data_path = '../data/source_datasets/CMeEE/CMeEE_dev.json'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1. 数据读入"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data_df = pd.read_json(train_data_path)\n",
    "dev_data_df = pd.read_json(dev_data_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_label(x):\n",
    "    \n",
    "    entities = []\n",
    "    for entity in x:\n",
    "        entity_ = {}\n",
    "        idx = list(range(entity['start_idx'], entity['end_idx']+1))\n",
    "        entity_['idx'] = idx\n",
    "        entity_['type'] = entity['type']\n",
    "        entity_['entity'] = entity['entity']\n",
    "        entities.append(entity_)\n",
    "    \n",
    "    return entities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data_df['label'] = train_data_df['entities'].apply(lambda x: get_label(x))\n",
    "dev_data_df['label'] = dev_data_df['entities'].apply(lambda x: get_label(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data_df = train_data_df.loc[:,['text', 'label']]\n",
    "train_data_df['label'] = train_data_df['label'].apply(lambda x: str(x))\n",
    "dev_data_df = dev_data_df.loc[:,['text', 'label']]\n",
    "dev_data_df['label'] = dev_data_df['label'].apply(lambda x: str(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ner_train_dataset = Dataset(train_data_df)\n",
    "ner_dev_dataset = Dataset(dev_data_df, categories=ner_train_dataset.categories)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2. 词典创建和生成分词器"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tokenizer = Tokenizer(vocab='nghuyong/ernie-1.0', max_seq_len=128)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3. ID化"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ner_train_dataset.convert_to_ids(tokenizer)\n",
    "ner_dev_dataset.convert_to_ids(tokenizer)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "### 二、模型构建"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1. 模型参数设置"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "config = W2NERBertConfig.from_pretrained('nghuyong/ernie-1.0',\n",
    "                                         num_labels=len(ner_train_dataset.cat2id))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2. 模型创建"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "dl_module = W2NERBert.from_pretrained('nghuyong/ernie-1.0',\n",
    "                                    config=config)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "### 三、任务构建"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 1. 任务参数和必要部件设定"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 设置运行次数\n",
    "num_epoches = 6\n",
    "batch_size = 8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = get_default_model_optimizer(dl_module)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 注意lr衰减轮次的设定\n",
    "t_total = len(ner_train_dataset) // batch_size * num_epoches\n",
    "scheduler = get_default_linear_schedule_with_warmup(optimizer, t_total, warmup_ratio=0.1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 2. 任务创建"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Task(dl_module, optimizer, 'ce', cude_device=0, scheduler=scheduler, grad_clip=5.0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3. 训练"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.fit(ner_train_dataset,\n",
    "          ner_dev_dataset,\n",
    "          # lr=3e-5,\n",
    "          epochs=1,\n",
    "          batch_size=batch_size\n",
    "         )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<br>\n",
    "\n",
    "### 四、生成提交数据"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "from tqdm import tqdm\n",
    "from ark_nlp.model.ner.w2ner_bert import Predictor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ner_predictor_instance = Predictor(model.module, tokenizer, ner_train_dataset.cat2id)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_df = pd.read_json('../data/source_datasets/CMeEE/CMeEE_test.json')\n",
    "\n",
    "submit = []\n",
    "for _text in tqdm(test_df['text'].to_list()):\n",
    "    \n",
    "    entities = ner_predictor_instance.predict_one_sample(_text)\n",
    "    \n",
    "    entities_ = []\n",
    "    for entity_ in entities:\n",
    "        entities_.append({\n",
    "                \"start_idx\": entity_['idx'][0],\n",
    "                \"end_idx\": entity_['idx'][-1],\n",
    "                \"entity\": entity_['entity'],\n",
    "                \"type\": entity_['entity'],\n",
    "            })\n",
    "    \n",
    "    submit.append({\n",
    "        'text': _text,\n",
    "        'entities': entities_\n",
    "    })"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "output_path = '../data/output_datasets/CMeEE_test.json'\n",
    "\n",
    "with open(output_path,'w', encoding='utf-8') as f:\n",
    "    f.write(json.dumps(submit, ensure_ascii=False))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [transformers]",
   "language": "python",
   "name": "transformers"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}