第N2周：中文文本分类-Pytorch实现

一、前言

🍨 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

● 难度：夯实基础⭐⭐
● 语言：Python3、Pytorch3
● 时间：4月23日-4月28日
🍺要求：
1、熟悉NLP的基础知识

二、准备工作

环境搭建
Python 3.8
pytorch == 1.8.1
torchtext == 0.9.1

三、数据预处理

1.加载数据

import torch
import torch.nn as nn
import os,PIL,pathlib,warnings

warnings.filterwarnings("ignore")             #忽略警告信息

# win10系统
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
device
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)
train_data.head()
# 构造数据集迭代器
def coustom_data_iter(texts, labels):
    for x, y in zip(texts, labels):
        yield x, y
        
train_iter = coustom_data_iter(train_data[0].values[:], train_data[1].values[:])

2.构建词典

from torchtext.data.utils import get_tokenizer
from torchtext.vocab import build_vocab_from_iterator
# conda install jieba -y
import jieba

# 中文分词方法
tokenizer = jieba.lcut

def 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
vocab(['我','想','看','和平','精英','上','战神','必备','技巧','的','游戏','视频'])
label_name = list(set(train_data[1].values[:]))
print(label_name)
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'))

3.生成数据批次和迭代器

from torch.utils.data import DataLoader

def 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)

三、模型构建

1. 搭建模型

from torch import nn

class 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 = 0.5
        self.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)

2. 初始化模型

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

3. 定义训练与评估函数

import time

def train(dataloader):
    model.train()  # 切换为训练模式
    total_acc, train_loss, total_count = 0, 0, 0
    log_interval = 50
    start_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loss
        total_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_time
            print('| 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, 0
            start_time = time.time()

def evaluate(dataloader):
    model.eval()  # 切换为测试模式
    total_acc, train_loss, total_count = 0, 0, 0

    with 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

四、训练模型

1. 拆分数据集并运行模型

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 training

criterion = 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_acc
    print('-' * 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)