3 FastText实现情感分析（pytorch）

在本节中，我们基于FastText算法实现一个模型论文，比之前的使用更少的参数，训练速度也明显加快

准备数据

FastText论文中的关键概念之一是它们计算输入句子的n-gram，并将它们附加到句子的末尾。这里我们使用bi-grams。

For example, in the sentence “how are you ?”, the bi-grams are: “how are”, “are you” and “you ?”.

generate_bigrams采用已经被分词的句子，计算bi-grams并将其附加到句子的末尾。

def generate_bigrams(x):
    n_grams = set(zip(*[x[i:] for i in range(2)]))#zip可以将两个列表组装为元祖输出
    for n_gram in n_grams:
        x.append(' '.join(n_gram))
    return x
generate_bigrams(['This', 'film', 'is', 'terrible'])

'''
['This', 'film', 'is', 'terrible', 'This film', 'film is', 'is terrible']
'''

使用Field定义处理数据的方式

import torch
from torchtext.legacy import data
from torchtext.legacy import datasets

SEED = 1234

torch.manual_seed(SEED)
torch.backends.cudnn.deterministic = True

TEXT = data.Field(tokenize = 'spacy',
                  tokenizer_language = 'en_core_web_sm',
                  preprocessing = generate_bigrams)

LABEL = data.LabelField(dtype = torch.float)

加载数据并拆分

import random

train_data, test_data = datasets.IMDB.splits(TEXT, LABEL)

train_data, valid_data = train_data.split(random_state = random.seed(SEED))

构建词汇表并加载预训练的word embedding

MAX_VOCAB_SIZE = 25_000

TEXT.build_vocab(train_data, 
                 max_size = MAX_VOCAB_SIZE, 
                 vectors = "glove.6B.100d", 
                 unk_init = torch.Tensor.normal_)

LABEL.build_vocab(train_data)

创建iterator

BATCH_SIZE = 64

device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

train_iterator, valid_iterator, test_iterator = data.BucketIterator.splits(
    (train_data, valid_data, test_data), 
    batch_size = BATCH_SIZE, 
    device = device)

创建模型

该模型的参数比以前的模型少得多，因为它只有2个具有任何参数的层，即嵌入层和线性层。 看不到RNN组件！

它首先使用Embedding层（蓝色）为每个单词计算单词嵌入，然后计算所有单词嵌入的平均值（粉红色）并将其馈送到Linear层（银色），仅此而已！

我们使用avg_pool2d函数来实现2维平均。你可能会以为使用2维池化很奇怪，我们的句子是1维的啊，为什么要用二维的。你可以将word embedding视为二维网格，其中单词沿一个轴，而Word embedding沿另一个轴。下图是word embeeding维度为5的示例句子，构成了一个[4*5]的张量

avg_pool2d使用大小为Embedded.shape [1]（即句子的长度）乘以1的过滤器。这在下图中以粉红色显示。

我们计算过滤器覆盖的所有元素的平均值，然后过滤器向右滑动，为句子中每个单词的嵌入值的下一列计算平均值。

每个过滤器位置给我们一个值，即所有覆盖元素的平均值。 过滤器覆盖所有嵌入尺寸后，我们得到一个[1x5]张量。 然后将该张量通过线性层以产生我们的预测。

import torch.nn as nn
import torch.nn.functional as F

class FastText(nn.Module):
    def __init__(self, vocab_size, embedding_dim, output_dim, pad_idx):
        
        super().__init__()
        
        self.embedding = nn.Embedding(vocab_size, embedding_dim, padding_idx=pad_idx)
        
        self.fc = nn.Linear(embedding_dim, output_dim)
        
    def forward(self, text):
        
        #text = [sent len, batch size]
        
        embedded = self.embedding(text)
                
        #embedded = [sent len, batch size, emb dim]
        
        embedded = embedded.permute(1, 0, 2)
        
        #embedded = [batch size, sent len, emb dim]
        
        pooled = F.avg_pool2d(embedded, (embedded.shape[1], 1)).squeeze(1) 
        
        #pooled = [batch size, embedding_dim]
                
        return self.fc(pooled)

和以前一样，我们将创建FastText类

INPUT_DIM = len(TEXT.vocab)
EMBEDDING_DIM = 100
OUTPUT_DIM = 1
PAD_IDX = TEXT.vocab.stoi[TEXT.pad_token]

model = FastText(INPUT_DIM, EMBEDDING_DIM, OUTPUT_DIM, PAD_IDX)

查看模型中的参数数量，我们发现与第一个笔记本中的标准RNN大致相同，而前一个模型的参数只有一半。

def count_parameters(model):
    return sum(p.numel() for p in model.parameters() if p.requires_grad)

print(f'The model has {count_parameters(model):,} trainable parameters')
'''
The model has 2,500,301 trainable parameters
'''

并将预训练的向量复制到嵌入层

pretrained_embeddings = TEXT.vocab.vectors

model.embedding.weight.data.copy_(pretrained_embeddings)

不要忘了将<unk>,<pad>初始权重归零

UNK_IDX = TEXT.vocab.stoi[TEXT.unk_token]

model.embedding.weight.data[UNK_IDX] = torch.zeros(EMBEDDING_DIM)
model.embedding.weight.data[PAD_IDX] = torch.zeros(EMBEDDING_DIM)

训练模型

步骤和之前的一样

import torch.optim as optim

optimizer = optim.Adam(model.parameters())
criterion = nn.BCEWithLogitsLoss()

model = model.to(device)
criterion = criterion.to(device)
def binary_accuracy(preds, y):
    """
    Returns accuracy per batch, i.e. if you get 8/10 right, this returns 0.8, NOT 8
    """

    #round predictions to the closest integer
    rounded_preds = torch.round(torch.sigmoid(preds))
    correct = (rounded_preds == y).float() #convert into float for division 
    acc = correct.sum() / len(correct)
    return acc

def train(model, iterator, optimizer, criterion):
    
    epoch_loss = 0
    epoch_acc = 0
    
    model.train()
    
    for batch in iterator:
        
        optimizer.zero_grad()
        
        predictions = model(batch.text).squeeze(1)
        
        loss = criterion(predictions, batch.label)
        
        acc = binary_accuracy(predictions, batch.label)
        
        loss.backward()
        
        optimizer.step()
        
        epoch_loss += loss.item()
        epoch_acc += acc.item()
        
    return epoch_loss / len(iterator), epoch_acc / len(iterator)

def evaluate(model, iterator, criterion):
    
    epoch_loss = 0
    epoch_acc = 0
    
    model.eval()
    
    with torch.no_grad():
    
        for batch in iterator:

            predictions = model(batch.text).squeeze(1)
            
            loss = criterion(predictions, batch.label)
            
            acc = binary_accuracy(predictions, batch.label)

            epoch_loss += loss.item()
            epoch_acc += acc.item()
        
    return epoch_loss / len(iterator), epoch_acc / len(iterator)

import time

def epoch_time(start_time, end_time):
    elapsed_time = end_time - start_time
    elapsed_mins = int(elapsed_time / 60)
    elapsed_secs = int(elapsed_time - (elapsed_mins * 60))
    return elapsed_mins, elapsed_secs

最后训练模型

N_EPOCHS = 5

best_valid_loss = float('inf')

for epoch in range(N_EPOCHS):

    start_time = time.time()
    
    train_loss, train_acc = train(model, train_iterator, optimizer, criterion)
    valid_loss, valid_acc = evaluate(model, valid_iterator, criterion)
    
    end_time = time.time()

    epoch_mins, epoch_secs = epoch_time(start_time, end_time)
    
    if valid_loss < best_valid_loss:
        best_valid_loss = valid_loss
        torch.save(model.state_dict(), 'tut3-model.pt')
    
    print(f'Epoch: {epoch+1:02} | Epoch Time: {epoch_mins}m {epoch_secs}s')
    print(f'\tTrain Loss: {train_loss:.3f} | Train Acc: {train_acc*100:.2f}%')
    print(f'\t Val. Loss: {valid_loss:.3f} |  Val. Acc: {valid_acc*100:.2f}%')

'''
Epoch: 01 | Epoch Time: 0m 7s
	Train Loss: 0.688 | Train Acc: 61.31%
	 Val. Loss: 0.637 |  Val. Acc: 72.46%
Epoch: 02 | Epoch Time: 0m 6s
	Train Loss: 0.651 | Train Acc: 75.04%
	 Val. Loss: 0.507 |  Val. Acc: 76.92%
Epoch: 03 | Epoch Time: 0m 6s
	Train Loss: 0.578 | Train Acc: 79.91%
	 Val. Loss: 0.424 |  Val. Acc: 80.97%
Epoch: 04 | Epoch Time: 0m 6s
	Train Loss: 0.501 | Train Acc: 83.97%
	 Val. Loss: 0.377 |  Val. Acc: 84.34%
Epoch: 05 | Epoch Time: 0m 6s
	Train Loss: 0.435 | Train Acc: 86.96%
	 Val. Loss: 0.363 |  Val. Acc: 86.18%
'''

在测试集上的表现如何呢

model.load_state_dict(torch.load('tut3-model.pt'))

test_loss, test_acc = evaluate(model, test_iterator, criterion)

print(f'Test Loss: {test_loss:.3f} | Test Acc: {test_acc*100:.2f}%')
'''
Test Loss: 0.381 | Test Acc: 85.42%
'''

用户输入

和之前的一样，先让句子产生bi-gram,然后通过Text.vocab.stoi找到对应token的值

import spacy
nlp = spacy.load('en_core_web_sm')

def predict_sentiment(model, sentence):
    model.eval()
    tokenized = generate_bigrams([tok.text for tok in nlp.tokenizer(sentence)])
    indexed = [TEXT.vocab.stoi[t] for t in tokenized]
    tensor = torch.LongTensor(indexed).to(device)
    tensor = tensor.unsqueeze(1)
    prediction = torch.sigmoid(model(tensor))
    return prediction.item()

'''
predict_sentiment(model, "This film is terrible")
predict_sentiment(model, "This film is great")
'''

下一步

将使用CNN用在情感分析任务上