第N5周：Pytorch文本分类入门

• 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

一、前期准备

1.环境安装

确保安装了torchtext与portalocker库

2.加载数据

import torch

# 强制使用 CPU
device = torch.device("cpu")
print(f"Forcing use of device: {device}")

# 确保模型和数据都使用 CPU
# model = model.to(device)
# data = data.to(device)

Forcing use of device: cpu

import torch 
import torch.nn as nn
import torchvision
from torchvision import transforms, datasets
import os, PIL, pathlib, warnings

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

from torchtext.datasets import AG_NEWS

train_iter = AG_NEWS(split='train')

3.构建词典

from torchtext.data.utils import get_tokenizer
from torchtext.vocab import build_vocab_from_iterator

tokenizer = get_tokenizer('basic_english') # 返回分词器函数

def yield_tokens(data_iter):
    for _, text in data_iter:
        yield tokenizer(text)

vocab = build_vocab_from_iterator(yield_tokens(train_iter), 
                                  specials=['']) 

vocab.set_default_index(vocab[''])  # 设置默认索引
vocab(['here', 'is', 'an', 'example'])

[475, 21, 30, 5297]

text_pipeline = lambda x: vocab(tokenizer(x))
label_pipeline = lambda x: int(x) - 1

text_pipeline('here is the an example')

[475, 21, 2, 30, 5297]

label_pipeline('10')

9

4.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 (_label, _text) 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中输入元素的累积和

    return label_list.to(device), text_list.to(device), offsets.to(device)  

# 数据加载器
data_loader = DataLoader(train_iter, 
                         batch_size=8, 
                         shuffle=False, 
                         collate_fn=collate_batch)

二、准备模型

1.定义模型

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)
        self.fc.weight.data.uniform_(-initrange, initrange)
        self.fc.bias.data.zero_()   

    def forward(self, text, offsets):
        embedded = self.embedding(text, offsets)
        return self.fc(embedded)

2.定义实例

num_class = len(set([label for (label, text) in train_iter]))
vocab_size = len(vocab) 
em_size = 64
model = TextClassificationModel(vocab_size, em_size, num_class).to(device)

3.定义训练函数与评估函数

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 = 500
    start_time = time.time()

    for idx, (label, text, 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
        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, (label, text, 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

三、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

# 然后运行您的代码

# 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

train_iter, test_iter = AG_NEWS() # 加载数据
train_dataset = to_map_style_dataset(train_iter)
test_dataset = to_map_style_dataset(test_iter)
num_train = int(len(train_dataset) * 0.95)

split_train_, split_valid_ = random_split(train_dataset, 
                                          [num_train, len(train_dataset) - num_train])

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)
test_dataloader = DataLoader(test_dataset, batch_size=BATCH_SIZE,
                             shuffle = True, collate_fn=collate_batch)   

for epoch in range(1, EPOCHS + 1):
    epoch_start_time = time.time()
    train(train_datasetloader)
    val_acc, val_loss = evaluate(valid_dataloader)

    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}'.format(epoch, time.time() - epoch_start_time, val_acc, val_loss))
    print('-' * 69)

| epoch 1 | 500/1782 batches| train_acc 0.901 train_loss 0.00458
| epoch 1 | 1000/1782 batches| train_acc 0.905 train_loss 0.00438
| epoch 1 | 1500/1782 batches| train_acc 0.908 train_loss 0.00437
---------------------------------------------------------------------
| epoch 1 | time:6.30s |valid_acc 0.907 | valid_loss 0.004
---------------------------------------------------------------------
| epoch 2 | 500/1782 batches| train_acc 0.917 train_loss 0.00381
| epoch 2 | 1000/1782 batches| train_acc 0.917 train_loss 0.00383
| epoch 2 | 1500/1782 batches| train_acc 0.917 train_loss 0.00386
---------------------------------------------------------------------
| epoch 2 | time:6.26s |valid_acc 0.911 | valid_loss 0.004
---------------------------------------------------------------------
| epoch 3 | 500/1782 batches| train_acc 0.929 train_loss 0.00330
| epoch 3 | 1000/1782 batches| train_acc 0.927 train_loss 0.00340
| epoch 3 | 1500/1782 batches| train_acc 0.923 train_loss 0.00354
---------------------------------------------------------------------
| epoch 3 | time:6.21s |valid_acc 0.935 | valid_loss 0.003
---------------------------------------------------------------------
| epoch 4 | 500/1782 batches| train_acc 0.933 train_loss 0.00306
| epoch 4 | 1000/1782 batches| train_acc 0.932 train_loss 0.00311
| epoch 4 | 1500/1782 batches| train_acc 0.929 train_loss 0.00318
---------------------------------------------------------------------
| epoch 4 | time:6.22s |valid_acc 0.916 | valid_loss 0.003
---------------------------------------------------------------------
| epoch 5 | 500/1782 batches| train_acc 0.948 train_loss 0.00253
| epoch 5 | 1000/1782 batches| train_acc 0.949 train_loss 0.00242
| epoch 5 | 1500/1782 batches| train_acc 0.951 train_loss 0.00238
---------------------------------------------------------------------
| epoch 5 | time:6.23s |valid_acc 0.954 | valid_loss 0.002
---------------------------------------------------------------------
| epoch 6 | 500/1782 batches| train_acc 0.951 train_loss 0.00241
| epoch 6 | 1000/1782 batches| train_acc 0.952 train_loss 0.00236
| epoch 6 | 1500/1782 batches| train_acc 0.952 train_loss 0.00235
---------------------------------------------------------------------
| epoch 6 | time:6.26s |valid_acc 0.954 | valid_loss 0.002
---------------------------------------------------------------------
| epoch 7 | 500/1782 batches| train_acc 0.954 train_loss 0.00228
| epoch 7 | 1000/1782 batches| train_acc 0.951 train_loss 0.00238
| epoch 7 | 1500/1782 batches| train_acc 0.954 train_loss 0.00228
---------------------------------------------------------------------
| epoch 7 | time:6.26s |valid_acc 0.954 | valid_loss 0.002
---------------------------------------------------------------------
| epoch 8 | 500/1782 batches| train_acc 0.953 train_loss 0.00227
| epoch 8 | 1000/1782 batches| train_acc 0.955 train_loss 0.00224
| epoch 8 | 1500/1782 batches| train_acc 0.954 train_loss 0.00224
---------------------------------------------------------------------
| epoch 8 | time:6.32s |valid_acc 0.954 | valid_loss 0.002
---------------------------------------------------------------------
| epoch 9 | 500/1782 batches| train_acc 0.955 train_loss 0.00218
| epoch 9 | 1000/1782 batches| train_acc 0.953 train_loss 0.00227
| epoch 9 | 1500/1782 batches| train_acc 0.955 train_loss 0.00227
---------------------------------------------------------------------
| epoch 9 | time:6.24s |valid_acc 0.954 | valid_loss 0.002
---------------------------------------------------------------------
| epoch 10 | 500/1782 batches| train_acc 0.952 train_loss 0.00229
| epoch 10 | 1000/1782 batches| train_acc 0.955 train_loss 0.00220
| epoch 10 | 1500/1782 batches| train_acc 0.956 train_loss 0.00220
---------------------------------------------------------------------
| epoch 10 | time:6.29s |valid_acc 0.954 | valid_loss 0.002
---------------------------------------------------------------------

2.使用测试数据集评估模型

print('Checking the results of test dataset.')  
test_acc, test_loss = evaluate(test_dataloader) 
print('test accuracy {:8.3f}'.format(test_acc))

Checking the results of test dataset.
test accuracy 0.910

四、总结
本周主要学习了文本分类，学习使用一个简单的模型来进行文本分类，同时了解到了self.embedding.weight.data.uniform_(-initrange, initrange)使用均匀分布的随机值来初始化权重，这种方法可以使模型在开始训练时有一定随机性，有助于避免梯度消失和梯度爆炸等问题。