【动手学深度学习-Pytorch版】BERT预测系列——用于预测的BERT数据集

本小节的主要任务即是将wiki数据集转成BERT输入序列，具体的任务包括：

• 读取wiki数据集
• 生成下一句预测任务的数据—>主要用于_get_nsp_data_from_paragraph函数
• 从输入paragraph生成用于下一句预测的训练样本：_get_nsp_data_from_paragraph
• 生成遮蔽语言模型任务的数据—>将生成的tokens的一部分随机换成masked的tokens，用于_get_mlm_data_from_tokens函数
• 得到掩蔽语言模型的数据：_get_mlm_data_from_tokens函数
• 将输入信息附加特殊词元< mask >
• 下载并生成WikiText-2数据集，并从中生成预训练样本:load_data_wiki函数
• 得到train_iter与vocab

"""较小的语料库WikiText-2"""
import os
import random
import torch
from d2l import torch as d2l
#@save
d2l.DATA_HUB['wikitext-2'] = ('https://s3.amazonaws.com/research.metamind.io/wikitext/''wikitext-2-v1.zip', '3c914d17d80b1459be871a5039ac23e752a53cbe')
"""仅使用句号作为分隔符来拆分句子"""
#@save
def _read_wiki(data_dir):file_name = os.path.join(data_dir, 'wiki.train.tokens')with open(file_name, 'r',encoding='utf-8') as f:lines = f.readlines()# 大写字母转换为小写字母paragraphs = [line.strip().lower().split(' . ')for line in lines if len(line.split(' . ')) >= 2]random.shuffle(paragraphs)return paragraphs#  生成下一句预测任务的数据--->用于：_get_nsp_data_from_paragraph函数
#@save
def _get_next_sentence(sentence, next_sentence, paragraphs):if random.random() < 0.5:is_next = Trueelse:# paragraphs是三重列表的嵌套next_sentence = random.choice(random.choice(paragraphs))is_next = Falsereturn sentence, next_sentence, is_next"""
下面的函数通过调用_get_next_sentence函数从输入paragraph生成用于下一句预测的训练样本。
这里paragraph是句子列表，其中每个句子都是词元列表。自变量max_len指定预训练期间的BERT输入序列的最大长度。
"""
#@save
def _get_nsp_data_from_paragraph(paragraph, paragraphs, vocab, max_len):nsp_data_from_paragraph = []"""nsp_data_from_paragraph中的每一个元素都是(tokens,segments,is_next)(词元，句子属性，是否是下一个句子)"""for i in range(len(paragraph) - 1):tokens_a, tokens_b, is_next = _get_next_sentence(paragraph[i], paragraph[i + 1], paragraphs)# 考虑1个'<cls>'词元和2个'<sep>'词元if len(tokens_a) + len(tokens_b) + 3 > max_len:continuetokens, segments = d2l.get_tokens_and_segments(tokens_a, tokens_b)nsp_data_from_paragraph.append((tokens, segments, is_next))return nsp_data_from_paragraph# 生成遮蔽语言模型任务的数据---》将生成的tokens的一部分随机换成masked的tokens
# -》》用于_get_mlm_data_from_tokens函数
"""
输入：
1、tokens：表示BERT输入序列的词元的列表
2、candidate_pred_positions：不包括特殊词元的BERT输入序列的词元索引的列表（特殊词元在遮蔽语言模型任务中不被预测）
3、num_mlm_preds：指示预测的数量（选择15%要预测的随机词元）
"""
#@save
def _replace_mlm_tokens(tokens, candidate_pred_positions, num_mlm_preds,vocab):# 为遮蔽语言模型的输入创建新的词元副本，其中输入可能包含替换的“<mask>”或随机词元mlm_input_tokens = [token for token in tokens]pred_positions_and_labels = []# 打乱后用于在遮蔽语言模型任务中获取15%的随机词元进行预测random.shuffle(candidate_pred_positions)for mlm_pred_position in candidate_pred_positions:# 如果生成的预测数量已经超过了最大的预测值 15% 就停止if len(pred_positions_and_labels) >= num_mlm_preds:breakmasked_token = None# 80%的时间：将词替换为“<mask>”词元if random.random() < 0.8:masked_token = '<mask>'else:# 10%的时间：保持词不变if random.random() < 0.5:masked_token = tokens[mlm_pred_position]# 10%的时间：用随机词替换该词else:masked_token = random.choice(vocab.idx_to_token)# 将masked的位置填入随机词元或保持不变或<mask>mlm_input_tokens[mlm_pred_position] = masked_tokenpred_positions_and_labels.append((mlm_pred_position, tokens[mlm_pred_position]))return mlm_input_tokens, pred_positions_and_labels"""
输入：BERT输入序列的tokens
输出：
1、输入词元的索引【词元已经被masked】
2、发生预测的词元索引
3、发生预测的标签索引
"""
"""当然，会有相关的词元会被masked"""
#@save
def _get_mlm_data_from_tokens(tokens, vocab):candidate_pred_positions = []# tokens是一个字符串列表for i, token in enumerate(tokens):# 在遮蔽语言模型任务中不会预测特殊词元if token in ['<cls>', '<sep>']:continuecandidate_pred_positions.append(i)# 遮蔽语言模型任务中预测15%的随机词元num_mlm_preds = max(1, round(len(tokens) * 0.15))mlm_input_tokens, pred_positions_and_labels = _replace_mlm_tokens(tokens, candidate_pred_positions, num_mlm_preds, vocab)pred_positions_and_labels = sorted(pred_positions_and_labels,key=lambda x: x[0])pred_positions = [v[0] for v in pred_positions_and_labels]mlm_pred_labels = [v[1] for v in pred_positions_and_labels]return vocab[mlm_input_tokens], pred_positions, vocab[mlm_pred_labels]"""
将特殊的“<mask>”词元附加到输入
"""
#@save
def _pad_bert_inputs(examples, max_len, vocab):max_num_mlm_preds = round(max_len * 0.15)all_token_ids, all_segments, valid_lens,  = [], [], []all_pred_positions, all_mlm_weights, all_mlm_labels = [], [], []nsp_labels = []for (token_ids, pred_positions, mlm_pred_label_ids, segments,is_next) in examples:# 如果长度不够会加入<pad>all_token_ids.append(torch.tensor(token_ids + [vocab['<pad>']] * (max_len - len(token_ids)), dtype=torch.long))# 而且所有的<pad>的segments都是0all_segments.append(torch.tensor(segments + [0] * (max_len - len(segments)), dtype=torch.long))# valid_lens不包括'<pad>'的计数 只是对token_ids计数，并不是对all_token_ids计数valid_lens.append(torch.tensor(len(token_ids), dtype=torch.float32))all_pred_positions.append(torch.tensor(pred_positions + [0] * (max_num_mlm_preds - len(pred_positions)), dtype=torch.long))# 填充词元的预测将通过乘以0权重在损失中过滤掉all_mlm_weights.append(torch.tensor([1.0] * len(mlm_pred_label_ids) + [0.0] * (max_num_mlm_preds - len(pred_positions)),dtype=torch.float32))all_mlm_labels.append(torch.tensor(mlm_pred_label_ids + [0] * (max_num_mlm_preds - len(mlm_pred_label_ids)), dtype=torch.long))nsp_labels.append(torch.tensor(is_next, dtype=torch.long))return (all_token_ids, all_segments, valid_lens, all_pred_positions,all_mlm_weights, all_mlm_labels, nsp_labels)#@save
class _WikiTextDataset(torch.utils.data.Dataset):def __init__(self, paragraphs, max_len):# 输入paragraphs[i]是代表段落的句子字符串列表；# 而输出paragraphs[i]是代表段落的句子列表，其中每个句子都是词元列表paragraphs = [d2l.tokenize(paragraph, token='word') for paragraph in paragraphs]sentences = [sentence for paragraph in paragraphsfor sentence in paragraph]self.vocab = d2l.Vocab(sentences, min_freq=5, reserved_tokens=['<pad>', '<mask>', '<cls>', '<sep>'])# 获取下一句子预测任务的数据examples = []for paragraph in paragraphs:examples.extend(_get_nsp_data_from_paragraph(paragraph, paragraphs, self.vocab, max_len))# 获取遮蔽语言模型任务的数据examples = [(_get_mlm_data_from_tokens(tokens, self.vocab)+ (segments, is_next))for tokens, segments, is_next in examples]# 填充输入(self.all_token_ids, self.all_segments, self.valid_lens,self.all_pred_positions, self.all_mlm_weights,self.all_mlm_labels, self.nsp_labels) = _pad_bert_inputs(examples, max_len, self.vocab)def __getitem__(self, idx):return (self.all_token_ids[idx], self.all_segments[idx],self.valid_lens[idx], self.all_pred_positions[idx],self.all_mlm_weights[idx], self.all_mlm_labels[idx],self.nsp_labels[idx])def __len__(self):return len(self.all_token_ids)"""下载并生成WikiText-2数据集，并从中生成预训练样本"""
#@save
def load_data_wiki(batch_size, max_len):"""加载WikiText-2数据集"""num_workers = d2l.get_dataloader_workers()data_dir = d2l.download_extract('wikitext-2', 'wikitext-2')paragraphs = _read_wiki(data_dir)train_set = _WikiTextDataset(paragraphs, max_len)train_iter = torch.utils.data.DataLoader(train_set, batch_size,shuffle=True, num_workers=num_workers)return train_iter, train_set.vocab
"""将批量大小设置为512，将BERT输入序列的最大长度设置为64，我们打印出小批量的BERT预训练样本的形状。"""
"""同时会有(64*0.15)的遮蔽语言模型需要预测的位置"""
batch_size, max_len = 512, 64
train_iter, vocab = load_data_wiki(batch_size, max_len)if __name__=='__main__':for (tokens_X, segments_X, valid_lens_x, pred_positions_X, mlm_weights_X,mlm_Y, nsp_y) in train_iter:print(tokens_X.shape, segments_X.shape, valid_lens_x.shape,pred_positions_X.shape, mlm_weights_X.shape, mlm_Y.shape,nsp_y.shape)break