深度学习笔记40_中文文本分类-Pytorch实现

🍨 本文为🔗365天深度学习训练营中的学习记录博客
🍖 原作者：K同学啊 | 接辅导、项目定制

一、我的环境

1.语言环境：Python 3.8

2.编译器：Pycharm

3.深度学习环境：

torch==1.12.1+cu113
torchvision==0.13.1+cu113

二、导入数据

import torch
import torch.nn as nn
import torchvision
from torchvision import transforms, datasets
import os,PIL,pathlib,warningswarnings.filterwarnings("ignore")             #忽略警告信息
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")import pandas as pd# 加载自定义中文数据
train_data = pd.read_csv('./data/train.csv', sep='\t', header=None)
print(train_data.head())

结果：

                       0              1
0      还有双鸭山到淮阴的汽车票吗13号的   Travel-Query
1                从这里怎么回家   Travel-Query
2       随便播放一首专辑阁楼里的佛里的歌     Music-Play
3              给看一下墓王之王嘛  FilmTele-Play
4  我想看挑战两把s686打突变团竞的游戏视频     Video-Play

三、构建词典

from torchtext.data.utils import get_tokenizer
from torchtext.vocab import build_vocab_from_iterator
import jieba# 中文分词方法
tokenizer = jieba.lcutdef yield_tokens(data_iter):for text,_ in data_iter:yield tokenizer(text)vocab = build_vocab_from_iterator(yield_tokens(train_iter), specials=["<unk>"])
vocab.set_default_index(vocab["<unk>"]) # 设置默认索引，如果找不到单词，则会选择默认索引print(vocab(['我','想','看','和平','精英','上','战神','必备','技巧','的','游戏','视频']))

结果：[2, 10, 13, 973, 1079, 146, 7724, 7574, 7793, 1, 186, 28]

text_pipeline  = lambda x: vocab(tokenizer(x))
label_pipeline = lambda x: label_name.index(x)print(text_pipeline('我想看和平精英上战神必备技巧的游戏视频'))
print(label_pipeline('Video-Play'))

结果：[2, 10, 13, 973, 1079, 146, 7724, 7574, 7793, 1, 186, 28]
4

四、生成数据批次和迭代器

from torch.utils.data import DataLoaderdef collate_batch(batch):label_list, text_list, offsets = [], [], [0]for (_text, _label) in batch:# 标签列表label_list.append(label_pipeline(_label))# 文本列表processed_text = torch.tensor(text_pipeline(_text), dtype=torch.int64)text_list.append(processed_text)# 偏移量，即语句的总词汇量offsets.append(processed_text.size(0))label_list = torch.tensor(label_list, dtype=torch.int64)text_list = torch.cat(text_list)offsets = torch.tensor(offsets[:-1]).cumsum(dim=0)  # 返回维度dim中输入元素的累计和return text_list.to(device), label_list.to(device), offsets.to(device)# 数据加载器，调用示例
dataloader = DataLoader(train_iter,batch_size=8,shuffle=False,collate_fn=collate_batch)

五、定义模型

from torch import nnclass TextClassificationModel(nn.Module):def __init__(self, vocab_size, embed_dim, num_class):super(TextClassificationModel, self).__init__()self.embedding = nn.EmbeddingBag(vocab_size,  # 词典大小embed_dim,  # 嵌入的维度sparse=False)  #self.fc = nn.Linear(embed_dim, num_class)self.init_weights()def init_weights(self):initrange = 0.5self.embedding.weight.data.uniform_(-initrange, initrange)  # 初始化权重self.fc.weight.data.uniform_(-initrange, initrange)self.fc.bias.data.zero_()  # 偏置值归零def forward(self, text, offsets):embedded = self.embedding(text, offsets)return self.fc(embedded)

六、定义实例

num_class  = len(label_name)
vocab_size = len(vocab)
em_size    = 64
model      = TextClassificationModel(vocab_size, em_size, num_class).to(device)

七、定义训练函数与评估函数

import timedef train(dataloader):model.train()  # 切换为训练模式total_acc, train_loss, total_count = 0, 0, 0log_interval = 50start_time = time.time()for idx, (text, label, offsets) in enumerate(dataloader):predicted_label = model(text, offsets)optimizer.zero_grad()  # grad属性归零loss = criterion(predicted_label, label)  # 计算网络输出和真实值之间的差距，label为真实值loss.backward()  # 反向传播torch.nn.utils.clip_grad_norm_(model.parameters(), 0.1)  # 梯度裁剪optimizer.step()  # 每一步自动更新# 记录acc与losstotal_acc += (predicted_label.argmax(1) == label).sum().item()train_loss += loss.item()total_count += label.size(0)if idx % log_interval == 0 and idx > 0:elapsed = time.time() - start_timeprint('| epoch {:1d} | {:4d}/{:4d} batches ''| train_acc {:4.3f} train_loss {:4.5f}'.format(epoch, idx, len(dataloader),total_acc / total_count, train_loss / total_count))total_acc, train_loss, total_count = 0, 0, 0start_time = time.time()def evaluate(dataloader):model.eval()  # 切换为测试模式total_acc, train_loss, total_count = 0, 0, 0with torch.no_grad():for idx, (text, label, offsets) in enumerate(dataloader):predicted_label = model(text, offsets)loss = criterion(predicted_label, label)  # 计算loss值# 记录测试数据total_acc += (predicted_label.argmax(1) == label).sum().item()train_loss += loss.item()total_count += label.size(0)return total_acc / total_count, train_loss / total_count

八、训练模型

from torch.utils.data.dataset import random_split
from torchtext.data.functional import to_map_style_dataset# 超参数
EPOCHS = 10  # epoch
LR = 5  # 学习率
BATCH_SIZE = 64  # batch size for trainingcriterion = torch.nn.CrossEntropyLoss()
optimizer = torch.optim.SGD(model.parameters(), lr=LR)
scheduler = torch.optim.lr_scheduler.StepLR(optimizer, 1.0, gamma=0.1)
total_accu = None# 构建数据集
train_iter = coustom_data_iter(train_data[0].values[:], train_data[1].values[:])
train_dataset = to_map_style_dataset(train_iter)split_train_, split_valid_ = random_split(train_dataset,[int(len(train_dataset) * 0.8), int(len(train_dataset) * 0.2)])train_dataloader = DataLoader(split_train_, batch_size=BATCH_SIZE,shuffle=True, collate_fn=collate_batch)valid_dataloader = DataLoader(split_valid_, batch_size=BATCH_SIZE,shuffle=True, collate_fn=collate_batch)for epoch in range(1, EPOCHS + 1):epoch_start_time = time.time()train(train_dataloader)val_acc, val_loss = evaluate(valid_dataloader)# 获取当前的学习率lr = optimizer.state_dict()['param_groups'][0]['lr']if total_accu is not None and total_accu > val_acc:scheduler.step()else:total_accu = val_accprint('-' * 69)print('| epoch {:1d} | time: {:4.2f}s | ''valid_acc {:4.3f} valid_loss {:4.3f} | lr {:4.6f}'.format(epoch,time.time() - epoch_start_time,val_acc, val_loss, lr))print('-' * 69)

结果：

Batch [50/152], Loss: 0.0340, Accuracy: 0.4203
Batch [100/152], Loss: 0.0235, Accuracy: 0.5851
Batch [150/152], Loss: 0.0309, Accuracy: 0.6572
---------------------------------------------------------------------
| epoch 1 | time: 0.55s | valid_acc 0.814 valid_loss 0.012 | lr 5.000000
---------------------------------------------------------------------
Batch [50/152], Loss: 0.0104, Accuracy: 0.8165
Batch [100/152], Loss: 0.0099, Accuracy: 0.8215
Batch [150/152], Loss: 0.0092, Accuracy: 0.8329
---------------------------------------------------------------------
| epoch 2 | time: 0.44s | valid_acc 0.855 valid_loss 0.008 | lr 5.000000
---------------------------------------------------------------------
Batch [50/152], Loss: 0.0068, Accuracy: 0.8790
Batch [100/152], Loss: 0.0065, Accuracy: 0.8778
Batch [150/152], Loss: 0.0064, Accuracy: 0.8809
---------------------------------------------------------------------
| epoch 3 | time: 0.44s | valid_acc 0.874 valid_loss 0.007 | lr 5.000000
---------------------------------------------------------------------
Batch [50/152], Loss: 0.0050, Accuracy: 0.9105
Batch [100/152], Loss: 0.0051, Accuracy: 0.9101
Batch [150/152], Loss: 0.0048, Accuracy: 0.9130
---------------------------------------------------------------------
| epoch 4 | time: 0.44s | valid_acc 0.882 valid_loss 0.006 | lr 5.000000
---------------------------------------------------------------------
Batch [50/152], Loss: 0.0039, Accuracy: 0.9366
Batch [100/152], Loss: 0.0039, Accuracy: 0.9339
Batch [150/152], Loss: 0.0038, Accuracy: 0.9350
---------------------------------------------------------------------
| epoch 5 | time: 0.44s | valid_acc 0.896 valid_loss 0.006 | lr 5.000000
---------------------------------------------------------------------
Batch [50/152], Loss: 0.0028, Accuracy: 0.9519
Batch [100/152], Loss: 0.0030, Accuracy: 0.9517
Batch [150/152], Loss: 0.0030, Accuracy: 0.9494
---------------------------------------------------------------------
| epoch 6 | time: 0.44s | valid_acc 0.898 valid_loss 0.005 | lr 5.000000
---------------------------------------------------------------------
Batch [50/152], Loss: 0.0025, Accuracy: 0.9580
Batch [100/152], Loss: 0.0024, Accuracy: 0.9616
Batch [150/152], Loss: 0.0024, Accuracy: 0.9609
---------------------------------------------------------------------
| epoch 7 | time: 0.44s | valid_acc 0.902 valid_loss 0.005 | lr 5.000000
---------------------------------------------------------------------
Batch [50/152], Loss: 0.0018, Accuracy: 0.9764
Batch [100/152], Loss: 0.0019, Accuracy: 0.9739
Batch [150/152], Loss: 0.0019, Accuracy: 0.9724
---------------------------------------------------------------------
| epoch 8 | time: 0.44s | valid_acc 0.900 valid_loss 0.005 | lr 5.000000
---------------------------------------------------------------------
Batch [50/152], Loss: 0.0015, Accuracy: 0.9810
Batch [100/152], Loss: 0.0014, Accuracy: 0.9817
Batch [150/152], Loss: 0.0014, Accuracy: 0.9818
---------------------------------------------------------------------
| epoch 9 | time: 0.49s | valid_acc 0.906 valid_loss 0.005 | lr 0.500000
---------------------------------------------------------------------
Batch [50/152], Loss: 0.0013, Accuracy: 0.9831
Batch [100/152], Loss: 0.0013, Accuracy: 0.9831
Batch [150/152], Loss: 0.0014, Accuracy: 0.9825
---------------------------------------------------------------------
| epoch 10 | time: 0.54s | valid_acc 0.906 valid_loss 0.005 | lr 0.500000
---------------------------------------------------------------------

九、预测

def predict(text, text_pipeline):with torch.no_grad():text = torch.tensor(text_pipeline(text))output = model(text, torch.tensor([0]))return output.argmax(1).item()# ex_text_str = "随便播放一首专辑阁楼里的佛里的歌"
ex_text_str = "还有双鸭山到淮阴的汽车票吗13号的"model = model.to("cpu")print("该文本的类别是：%s" %label_name[predict(ex_text_str, text_pipeline)])

该文本的类别是：Travel-Query

总结：

‌语料库（原始文本）‌：
来源包括维基百科、网页文本、新闻资讯及内部文本。
‌文本清洗‌：
清洗原始文本，包括去除标点符号和特殊字符。该流程主要用于将原始文本数据转化为可用于模型训练的数值化向量，再通过深度学习模型进行文本分类。
- ‌分词‌：
  使用jieba分词工具对清洗后的文本进行分词处理。
- ‌建模‌：
  采用不同的模型进行文本建模，包括循环神经网络（RNN）、卷积神经网络（CNN）、门控循环单元（GRU）和长短期记忆网络（LSTM）。
- ‌文本向量化‌：
  将分词后的文本转换为向量表示，方法包括TF-IDF和Word2vec。