You need to sign in or sign up before continuing.
bert_train.py 15.2 KB
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413
# -*- coding: utf-8 -*-
"""
BERT情感分析模型训练脚本
"""
import argparse
import os
import pandas as pd
import torch
import torch.nn as nn
from torch.utils.data import Dataset, DataLoader
from transformers import BertTokenizer, BertModel
from sklearn.metrics import accuracy_score, f1_score, classification_report, roc_auc_score
from typing import List, Tuple
import warnings
import requests
from pathlib import Path

from base_model import BaseModel
from utils import load_corpus_bert

# 忽略transformers的警告
warnings.filterwarnings("ignore")
os.environ["KMP_DUPLICATE_LIB_OK"] = "TRUE"


class BertDataset(Dataset):
    """BERT数据集"""
    
    def __init__(self, data: List[Tuple[str, int]]):
        self.data = [item[0] for item in data]
        self.labels = [item[1] for item in data]
    
    def __getitem__(self, index):
        return self.data[index], self.labels[index]
    
    def __len__(self):
        return len(self.labels)


class BertClassifier(nn.Module):
    """BERT分类器网络"""
    
    def __init__(self, input_size):
        super(BertClassifier, self).__init__()
        self.fc = nn.Linear(input_size, 1)
        self.sigmoid = nn.Sigmoid()
    
    def forward(self, x):
        out = self.fc(x)
        out = self.sigmoid(out)
        return out


class BertModel_Custom(BaseModel):
    """BERT情感分析模型"""
    
    def __init__(self, model_path: str = "./model/chinese_wwm_pytorch"):
        super().__init__("BERT")
        self.model_path = model_path
        self.tokenizer = None
        self.bert = None
        self.classifier = None
        self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
        
    def _download_bert_model(self):
        """自动下载BERT预训练模型"""
        print(f"BERT模型不存在,正在下载中文BERT预训练模型...")
        print("下载来源: bert-base-chinese (Hugging Face)")
        
        try:
            # 创建模型目录
            os.makedirs(self.model_path, exist_ok=True)
            
            # 使用Hugging Face的中文BERT模型
            model_name = "bert-base-chinese"
            print(f"正在从Hugging Face下载 {model_name}...")
            
            # 下载tokenizer
            print("下载分词器...")
            tokenizer = BertTokenizer.from_pretrained(model_name)
            tokenizer.save_pretrained(self.model_path)
            
            # 下载模型
            print("下载BERT模型...")
            bert_model = BertModel.from_pretrained(model_name)
            bert_model.save_pretrained(self.model_path)
            
            print(f"✅ BERT模型下载完成,保存在: {self.model_path}")
            return True
            
        except Exception as e:
            print(f"❌ BERT模型下载失败: {e}")
            print("\n💡 您可以手动下载BERT模型:")
            print("1. 访问 https://huggingface.co/bert-base-chinese")
            print("2. 或使用哈工大中文BERT: https://github.com/ymcui/Chinese-BERT-wwm")
            print(f"3. 将模型文件解压到: {self.model_path}")
            return False
    
    def _load_bert(self):
        """加载BERT模型和分词器"""
        print(f"加载BERT模型: {self.model_path}")
        
        # 如果模型不存在,尝试自动下载
        if not os.path.exists(self.model_path) or not any(os.scandir(self.model_path)):
            print("BERT模型不存在,尝试自动下载...")
            if not self._download_bert_model():
                raise FileNotFoundError(f"BERT模型下载失败,请手动下载到: {self.model_path}")
        
        try:
            self.tokenizer = BertTokenizer.from_pretrained(self.model_path)
            self.bert = BertModel.from_pretrained(self.model_path).to(self.device)
            
            # 冻结BERT参数
            for param in self.bert.parameters():
                param.requires_grad = False
                
            print("✅ BERT模型加载完成")
            
        except Exception as e:
            print(f"❌ BERT模型加载失败: {e}")
            print("尝试使用在线模型...")
            
            # 如果本地加载失败,尝试直接使用在线模型
            try:
                model_name = "bert-base-chinese"
                self.tokenizer = BertTokenizer.from_pretrained(model_name)
                self.bert = BertModel.from_pretrained(model_name).to(self.device)
                
                # 冻结BERT参数
                for param in self.bert.parameters():
                    param.requires_grad = False
                    
                print("✅ 在线BERT模型加载完成")
                
            except Exception as e2:
                print(f"❌ 在线模型也加载失败: {e2}")
                raise FileNotFoundError(f"无法加载BERT模型,请检查网络连接或手动下载模型到: {self.model_path}")
    
    def train(self, train_data: List[Tuple[str, int]], **kwargs) -> None:
        """训练BERT模型"""
        print(f"开始训练 {self.model_name} 模型...")
        
        # 加载BERT
        self._load_bert()
        
        # 超参数
        learning_rate = kwargs.get('learning_rate', 1e-3)
        num_epochs = kwargs.get('num_epochs', 10)
        batch_size = kwargs.get('batch_size', 100)
        input_size = kwargs.get('input_size', 768)
        decay_rate = kwargs.get('decay_rate', 0.9)
        
        print(f"BERT超参数: lr={learning_rate}, epochs={num_epochs}, "
              f"batch_size={batch_size}, input_size={input_size}")
        
        # 创建数据集
        train_dataset = BertDataset(train_data)
        train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)
        
        # 创建分类器
        self.classifier = BertClassifier(input_size).to(self.device)
        
        # 损失函数和优化器
        criterion = nn.BCELoss()
        optimizer = torch.optim.Adam(self.classifier.parameters(), lr=learning_rate)
        scheduler = torch.optim.lr_scheduler.ExponentialLR(optimizer, gamma=decay_rate)
        
        # 训练循环
        self.bert.eval()  # BERT始终保持评估模式
        self.classifier.train()
        
        for epoch in range(num_epochs):
            total_loss = 0
            num_batches = 0
            
            for i, (words, labels) in enumerate(train_loader):
                # 分词和编码
                tokens = self.tokenizer(words, padding=True, truncation=True, 
                                      max_length=512, return_tensors='pt')
                input_ids = tokens["input_ids"].to(self.device)
                attention_mask = tokens["attention_mask"].to(self.device)
                labels = torch.tensor(labels, dtype=torch.float32).to(self.device)
                
                # 获取BERT输出(冻结参数)
                with torch.no_grad():
                    bert_outputs = self.bert(input_ids, attention_mask=attention_mask)
                    bert_output = bert_outputs[0][:, 0]  # [CLS] token的输出
                
                # 分类器前向传播
                optimizer.zero_grad()
                outputs = self.classifier(bert_output)
                logits = outputs.view(-1)
                loss = criterion(logits, labels)
                
                # 反向传播
                loss.backward()
                optimizer.step()
                
                total_loss += loss.item()
                num_batches += 1
                
                if (i + 1) % 10 == 0:
                    avg_loss = total_loss / num_batches
                    print(f"Epoch [{epoch+1}/{num_epochs}], Step [{i+1}], Loss: {avg_loss:.4f}")
                    total_loss = 0
                    num_batches = 0
            
            # 学习率衰减
            scheduler.step()
            
            # 保存每个epoch的模型
            if kwargs.get('save_each_epoch', False):
                epoch_model_path = f"./model/bert_epoch_{epoch+1}.pth"
                os.makedirs(os.path.dirname(epoch_model_path), exist_ok=True)
                torch.save(self.classifier.state_dict(), epoch_model_path)
                print(f"已保存模型: {epoch_model_path}")
        
        self.is_trained = True
        print(f"{self.model_name} 模型训练完成!")
    
    def predict(self, texts: List[str]) -> List[int]:
        """预测文本情感"""
        if not self.is_trained:
            raise ValueError(f"模型 {self.model_name} 尚未训练,请先调用train方法")
        
        predictions = []
        batch_size = 32
        
        self.bert.eval()
        self.classifier.eval()
        
        with torch.no_grad():
            for i in range(0, len(texts), batch_size):
                batch_texts = texts[i:i+batch_size]
                
                # 分词和编码
                tokens = self.tokenizer(batch_texts, padding=True, truncation=True,
                                      max_length=512, return_tensors='pt')
                input_ids = tokens["input_ids"].to(self.device)
                attention_mask = tokens["attention_mask"].to(self.device)
                
                # 获取BERT输出
                bert_outputs = self.bert(input_ids, attention_mask=attention_mask)
                bert_output = bert_outputs[0][:, 0]
                
                # 分类器预测
                outputs = self.classifier(bert_output)
                outputs = outputs.view(-1)
                
                # 转换为类别标签
                preds = (outputs > 0.5).cpu().numpy()
                predictions.extend(preds.astype(int).tolist())
        
        return predictions
    
    def predict_single(self, text: str) -> Tuple[int, float]:
        """预测单条文本的情感"""
        if not self.is_trained:
            raise ValueError(f"模型 {self.model_name} 尚未训练,请先调用train方法")
        
        self.bert.eval()
        self.classifier.eval()
        
        with torch.no_grad():
            # 分词和编码
            tokens = self.tokenizer([text], padding=True, truncation=True,
                                  max_length=512, return_tensors='pt')
            input_ids = tokens["input_ids"].to(self.device)
            attention_mask = tokens["attention_mask"].to(self.device)
            
            # 获取BERT输出
            bert_outputs = self.bert(input_ids, attention_mask=attention_mask)
            bert_output = bert_outputs[0][:, 0]
            
            # 分类器预测
            output = self.classifier(bert_output)
            prob = output.item()
            
            prediction = int(prob > 0.5)
            confidence = prob if prediction == 1 else 1 - prob
        
        return prediction, confidence
    
    def save_model(self, model_path: str = None) -> None:
        """保存模型"""
        if not self.is_trained:
            raise ValueError(f"模型 {self.model_name} 尚未训练,无法保存")
        
        if model_path is None:
            model_path = f"./model/{self.model_name.lower()}_model.pth"
        
        os.makedirs(os.path.dirname(model_path), exist_ok=True)
        
        # 保存分类器和相关信息
        model_data = {
            'classifier_state_dict': self.classifier.state_dict(),
            'model_path': self.model_path,
            'input_size': 768,
            'device': str(self.device)
        }
        
        torch.save(model_data, model_path)
        print(f"模型已保存到: {model_path}")
    
    def load_model(self, model_path: str) -> None:
        """加载模型"""
        if not os.path.exists(model_path):
            raise FileNotFoundError(f"模型文件不存在: {model_path}")
        
        model_data = torch.load(model_path, map_location=self.device)
        
        # 设置BERT模型路径
        self.model_path = model_data['model_path']
        
        # 加载BERT
        self._load_bert()
        
        # 重建分类器
        input_size = model_data['input_size']
        self.classifier = BertClassifier(input_size).to(self.device)
        
        # 加载分类器权重
        self.classifier.load_state_dict(model_data['classifier_state_dict'])
        
        self.is_trained = True
        print(f"已加载模型: {model_path}")
    
    @staticmethod
    def load_data(train_path: str, test_path: str) -> Tuple[List[Tuple[str, int]], List[Tuple[str, int]]]:
        """加载BERT格式的数据"""
        print("加载训练数据...")
        train_data = load_corpus_bert(train_path)
        print(f"训练数据量: {len(train_data)}")
        
        print("加载测试数据...")
        test_data = load_corpus_bert(test_path)
        print(f"测试数据量: {len(test_data)}")
        
        return train_data, test_data


def main():
    """主函数"""
    parser = argparse.ArgumentParser(description='BERT情感分析模型训练')
    parser.add_argument('--train_path', type=str, default='./data/weibo2018/train.txt',
                        help='训练数据路径')
    parser.add_argument('--test_path', type=str, default='./data/weibo2018/test.txt',
                        help='测试数据路径')
    parser.add_argument('--model_path', type=str, default='./model/bert_model.pth',
                        help='模型保存路径')
    parser.add_argument('--bert_path', type=str, default='./model/chinese_wwm_pytorch',
                        help='BERT预训练模型路径')
    parser.add_argument('--epochs', type=int, default=10,
                        help='训练轮数')
    parser.add_argument('--batch_size', type=int, default=100,
                        help='批大小')
    parser.add_argument('--learning_rate', type=float, default=1e-3,
                        help='学习率')
    parser.add_argument('--eval_only', action='store_true',
                        help='仅评估已有模型,不进行训练')
    
    args = parser.parse_args()
    
    # 创建模型
    model = BertModel_Custom(args.bert_path)
    
    if args.eval_only:
        # 仅评估模式
        print("评估模式:加载已有模型进行评估")
        model.load_model(args.model_path)
        
        # 加载测试数据
        _, test_data = model.load_data(args.train_path, args.test_path)
        
        # 评估模型
        model.evaluate(test_data)
    else:
        # 训练模式
        # 加载数据
        train_data, test_data = model.load_data(args.train_path, args.test_path)
        
        # 训练模型
        model.train(
            train_data,
            num_epochs=args.epochs,
            batch_size=args.batch_size,
            learning_rate=args.learning_rate
        )
        
        # 评估模型
        model.evaluate(test_data)
        
        # 保存模型
        model.save_model(args.model_path)
        
        # 示例预测
        print("\n示例预测:")
        test_texts = [
            "今天天气真好,心情很棒",
            "这部电影太无聊了,浪费时间",
            "哈哈哈,太有趣了"
        ]
        
        for text in test_texts:
            pred, conf = model.predict_single(text)
            sentiment = "正面" if pred == 1 else "负面"
            print(f"文本: {text}")
            print(f"预测: {sentiment} (置信度: {conf:.4f})")
            print()


if __name__ == "__main__":
    main()