python代码实现了一个基于双向LSTM（Long Short-Term Memory）的情感分类模型，用于对IMDB影评数据进行情感分析，判断影评是正面还是负面

import torch
import torch.nn as nn
import torch.optim as optim
from torch.utils.data import DataLoader
from torchtext.vocab import GloVe
from torchtext.data.utils import get_tokenizer
import torchtext
from torchtext.vocab import build_vocab_from_iterator


# 使用torchtext加载IMDB数据集
from torchtext.datasets import IMDB

# 使用torchtext的tokenizer（basic_english分词器）
tokenizer = get_tokenizer('basic_english')

# 词汇表构建
specials = ["<unk>", "<pad>", "<bos>", "<eos>"]
def yield_tokens(sentences):
    for sentence in sentences:
        yield tokenizer(sentence)

# 下载并加载IMDB数据集
train_iter, test_iter = IMDB(split=('train', 'test'))

# 构建词汇表
vocab = build_vocab_from_iterator(yield_tokens([text for (text, label) in train_iter]), specials=specials, min_freq=1)
vocab.set_default_index(vocab["<unk>"])

# 加载GloVe词向量
glove = GloVe(name='6B', dim=100)
vocab.set_vectors(glove.stoi, glove.vectors, glove.dim)

# 定义文本处理函数，将文本转化为整数索引
def text_to_indices(texts, vocab, tokenizer):
    return [
        [vocab[token] if token in vocab else vocab["<unk>"] for token in tokenizer(text)]
        for text in texts
    ]

# 从训练集和测试集提取文本和标签
train_sentences, train_labels = zip(*[(text, label) for text, label in train_iter])
test_sentences,