BERT学习笔记：create_pretraining_data.py 运行及代码解读

1 简介

本文主要是自己理解的记录，方便以后回顾时不是从头开始看，如果你有幸看到这篇文章并且对你有些许帮助，我很荣幸，如果没有帮助也感谢你的浏览。

2 运行

首先介绍怎么运行，边运行边查看输出可以方便理解代码，创建tmp文件夹，加入以下三个文件：text.txt，output.txt，vocab.txt，是运行时必须指定的参数。

text.txt中为训练文本，随便找的文本

写诗做文章都是千古之事，而其中甘苦得失只有作者自己心里知道。
各位作者都是不同地位的人，怎么会空有虚名呢？

这两句话表达杜甫晚年对诗歌创作的见解，带有总结性质。
上句“千古事”是指留传久远，关系重大，下句“寸心知”是说对于文章，作者本人的理解感知最为明白。
这两句诗虽是以议论入诗，但对仗工整，语言高度概括，而且切中肯綮，含蕴丰富，很有哲理性。

vocab.txt 可以从Google发布的中文模型中解压缩找到。

output.txt为必须制定的输出目录

python create_pretraining_data.py --input_file=tmp/text.txt --output_file=tmp/output.txt --vocab_file=tmp/vocab.txt

 2 源码解释

2.1必须参数

flags.DEFINE_string("input_file", None,
                    "Input raw text file (or comma-separated list of files).")

flags.DEFINE_string(
    "output_file", None,
    "Output TF example file (or comma-separated list of files).")

flags.DEFINE_string("vocab_file", None,
                    "The vocabulary file that the BERT model was trained on.")

• 文件输入路径
• 输出文件路径
• 词典文件路径 (预训练好模型中能够找到)

 2.2可选参数

flags.DEFINE_bool(
    "do_lower_case", True,
    "Whether to lower case the input text. Should be True for uncased "
    "models and False for cased models.")

flags.DEFINE_integer("max_seq_length", 128, "Maximum sequence length.")

flags.DEFINE_integer("max_predictions_per_seq", 20,
                     "Maximum number of masked LM predictions per sequence.")

• 是否小写输入， 默认True
• 最大句子的长度
• 每一句MLM预测的百分比

flags.DEFINE_integer("random_seed", 12345, "Random seed for data generation.")

flags.DEFINE_integer(
    "dupe_factor", 10,
    "Number of times to duplicate the input data (with different masks).")

flags.DEFINE_float("masked_lm_prob", 0.15, "Masked LM probability.")

flags.DEFINE_float(
    "short_seq_prob", 0.1,
    "Probability of creating sequences which are shorter than the "
    "maximum length.")

• 用于数据生成的随机种子
• 复制输入数据的次数(使用不同的掩码) 默认循环10次
• Masked LM的比例

3 训练实例

3.1单独的实例训练

class TrainingInstance(object):
  """A single training instance (sentence pair)."""
  # 初始化
  def __init__(self, tokens, segment_ids, masked_lm_positions, masked_lm_labels,
               is_random_next):
    self.tokens = tokens
    self.segment_ids = segment_ids
    self.is_random_next = is_random_next
    self.masked_lm_positions = masked_lm_positions
    self.masked_lm_labels = masked_lm_labels
  # 字符串化
  def __str__(self):
    s = ""
    s += "tokens: %s\n" % (" ".join(
        [tokenization.printable_text(x) for x in self.tokens]))
    s += "segment_ids: %s\n" % (" ".join([str(x) for x in self.segment_ids]))
    s += "is_random_next: %s\n" % self.is_random_next
    s += "masked_lm_positions: %s\n" % (" ".join(
        [str(x) for x in self.masked_lm_positions]))
    s += "masked_lm_labels: %s\n" % (" ".join(
        [tokenization.printable_text(x) for x in self.masked_lm_labels]))
    s += "\n"
    return s

  def __repr__(self):
    return self.__str__()

 3.2 写训练实例到文件

def write_instance_to_example_files(instances, tokenizer, max_seq_length,
                                    max_predictions_per_seq, output_files):
  """Create TF example files from `TrainingInstance`s."""
  writers = []
  for output_file in output_files:
    writers.append(tf.python_io.TFRecordWriter(output_file))

  writer_index = 0

  total_written = 0
  for (inst_index, instance) in enumerate(instances):
    # 将字通过vocan.txt 转换为id 具体看tokenizer.py
    input_ids = tokenizer.convert_tokens_to_ids(instance.tokens)
    input_mask = [1] * len(input_ids)
    segment_ids = list(instance.segment_ids)
    assert len(input_ids) <= max_seq_length
    # 小于最大序列长度补0
    while len(input_ids) < max_seq_length:
      input_ids.append(0)
      input_mask.append(0)
      segment_ids.append(0)

    assert len(input_ids) == max_seq_length
    assert len(input_mask) == max_seq_length
    assert len(segment_ids) == max_seq_length

    masked_lm_positions = list(instance.masked_lm_positions)
    masked_lm_ids = tokenizer.convert_tokens_to_ids(instance.masked_lm_labels)
    masked_lm_weights = [1.0] * len(masked_lm_ids)

    while len(masked_lm_positions) < max_predictions_per_seq:
      masked_lm_positions.append(0)
      masked_lm_ids.append(0)
      masked_lm_weights.append(0.0)

    next_sentence_label = 1 if instance.is_random_next else 0

    features = collections.OrderedDict()
    features["input_ids"] = create_int_feature(input_ids)
    features["input_mask"] = create_int_feature(input_mask)
    features["segment_ids"] = create_int_feature(segment_ids)
    features["masked_lm_positions"] = create_int_feature(masked_lm_positions)
    features["masked_lm_ids"] = create_int_feature(masked_lm_ids)
    features["masked_lm_weights"] = create_float_feature(masked_lm_weights)
    features["next_sentence_labels"] = create_int_feature([next_sentence_label])
    tf_example = tf.train.Example(features=tf.train.Features(feature=features))

    writers[writer_index].write(tf_example.SerializeToString())
    writer_index = (writer_index + 1) % len(writers)

    total_written += 1

    if inst_index < 20:
      tf.logging.info("*** Example ***")
      tf.logging.info("tokens: %s" % " ".join(
          [tokenization.printable_text(x) for x in instance.tokens]))

      for feature_name in features.keys():
        feature = features[feature_name]
        values = []
        if feature.int64_list.value:
          values = feature.int64_list.value
        elif feature.float_list.value:
          values = feature.float_list.value
        tf.logging.info(
            "%s: %s" % (feature_name, " ".join([str(x) for x in values])))

  for writer in writers:
    writer.close()

  tf.logging.info("Wrote %d total instances", total_written)

 把格式好的用于训练的数据写入tfrecord文件中，这个方法的主要作用就是把已经处理好的数据写入二进制文件中

def create_int_feature(values):
  feature = tf.train.Feature(int64_list=tf.train.Int64List(value=list(values)))
  return feature


def create_float_feature(values):
  feature = tf.train.Feature(float_list=tf.train.FloatList(value=list(values)))
  return feature

创建不同的训练特征，主要将预处理好的数据转为Tensorflow中的格式

3.3 创建训练实例

def create_training_instances(input_files, tokenizer, max_seq_length,
                              dupe_factor, short_seq_prob, masked_lm_prob,
                              max_predictions_per_seq, rng):
  """Create `TrainingInstance`s from raw text."""
  all_documents = [[]]

  # Input file format:
  # (1) One sentence per line. These should ideally be actual sentences, not
  # entire paragraphs or arbitrary spans of text. (Because we use the
  # sentence boundaries for the "next sentence prediction" task).
  # (2) Blank lines between documents. Document boundaries are needed so
  # that the "next sentence prediction" task doesn't span between documents.
  # 每行是一个文档中的一句话，凭借这个来进行下一句预测的任务
  # 文档之间间隔一个空行, 因为下一个预测不跨文档 
  # all_documents最后形成三维数组, 每句形成一个一维数组, 每个段落中的每句放在一维中, 每个段落放 
  # 在一维中
  for input_file in input_files:
    with tf.gfile.GFile(input_file, "r") as reader:
      while True:
        line = tokenization.convert_to_unicode(reader.readline())
        if not line:
          break
        line = line.strip()

        # Empty lines are used as document delimiters
        if not line:
          all_documents.append([])
        tokens = tokenizer.tokenize(line)
        if tokens:
          all_documents[-1].append(tokens)

  # Remove empty documents
  all_documents = [x for x in all_documents if x]
  rng.shuffle(all_documents)

  vocab_words = list(tokenizer.vocab.keys())
  instances = []
  # dupe_factor默认值是10, 复制输入数据的次数, 相同的数据形成不同的掩码, 增加实例用于训练
  for _ in range(dupe_factor):
    for document_index in range(len(all_documents)):
      instances.extend(
          create_instances_from_document(
              all_documents, document_index, max_seq_length, short_seq_prob,
              masked_lm_prob, max_predictions_per_seq, vocab_words, rng))

  rng.shuffle(instances)
  print('instances', len(instances))
  return instances

 3.4根据但文档创建实例

def create_instances_from_document(
    all_documents, document_index, max_seq_length, short_seq_prob,
    masked_lm_prob, max_predictions_per_seq, vocab_words, rng):
  """Creates `TrainingInstance`s for a single document."""
  # document 相当于单个段落
  document = all_documents[document_index]

  # Account for [CLS], [SEP], [SEP]
  # 为了加入 [CLS], [SEP], [SEP] 所以长度减3
  max_num_tokens = max_seq_length - 3

  # 大部分情况下，我们希望把长度填充到最大长度，但是一少部分情况下我们希望采用短句来最小化预训练 
  # 和微调的差异。总的来说 target_seq_length 是一个粗略的目标，而 max_seq_length是一个强制的限制。
  target_seq_length = max_num_tokens
  if rng.random() < short_seq_prob:
    target_seq_length = rng.randint(2, max_num_tokens) #生成随机数
  
  instances = []
  current_chunk = []
  current_length = 0
  i = 0
  while i < len(document):
    segment = document[i] # segment是一个句子列表
    current_chunk.append(segment)
    current_length += len(segment)
    # i == len(document) - 1 为了将本段落的所有segment 都加入current_chunk中再往下执行
    if i == len(document) - 1 or current_length >= target_seq_length:
      if current_chunk:
        # `a_end` is how many segments from `current_chunk` go into the `A`
        # (first) sentence.
        a_end = 1
        if len(current_chunk) >= 2:
          # 随机生成1到len(current_chunk)-1的随机数
          a_end = rng.randint(1, len(current_chunk) - 1)

        # tokens_a 表示将文档中前a_end 个句子加入tokens_a中
        tokens_a = []
        for j in range(a_end):
          tokens_a.extend(current_chunk[j])

        # tokens_b 如果当前文档(段落)如果只有一个句子，随机填充一个句子作为下一个句子预测
        # 如果不是一个句子将 a_end到len(current_chunk) - 1 加到tokens_b中
        # 还有随机概率小于0.5的情况
        tokens_b = []
        # Random next
        # 如果is_random_next为True 训练时转化为0, False 转化为1
        is_random_next = False
        # 如果只有一个句子或者随机概率小于0.5 随机填充下一个句子
        if len(current_chunk) == 1 or rng.random() < 0.5:
          is_random_next = True
          target_b_length = target_seq_length - len(tokens_a)
          # 从总文档中随机填充一个
          for _ in range(10):
            random_document_index = rng.randint(0, len(all_documents) - 1)
            if random_document_index != document_index:
              break

          random_document = all_documents[random_document_index]
          random_start = rng.randint(0, len(random_document) - 1)
          for j in range(random_start, len(random_document)):
            tokens_b.extend(random_document[j])
            if len(tokens_b) >= target_b_length:
              break
          # We didn't actually use these segments so we "put them back" so
          # they don't go to waste.
          num_unused_segments = len(current_chunk) - a_end
          i -= num_unused_segments
        # Actual next
        else:
          is_random_next = False
          for j in range(a_end, len(current_chunk)):
            tokens_b.extend(current_chunk[j])
        # 将一对序列截断到最大序列长度， tokens_a, tokens_b 是当前所有文档总数(文档是当前段落)
        truncate_seq_pair(tokens_a, tokens_b, max_num_tokens, rng)

        assert len(tokens_a) >= 1
        assert len(tokens_b) >= 1
        # 下边几行看过论文的很好理解
        tokens = []
        segment_ids = []
        tokens.append("[CLS]")
        segment_ids.append(0)
        for token in tokens_a:
          tokens.append(token)
          segment_ids.append(0)

        tokens.append("[SEP]")
        segment_ids.append(0)

        for token in tokens_b:
          tokens.append(token)
          segment_ids.append(1)
        tokens.append("[SEP]")
        segment_ids.append(1)
        # tokens [CLS] 各 位 作 者 都 是 不 同 地 位 [MASK] 人 ， 怎 么 会 空 有 虚 名 呢 ？ [SEP] 这 两 句 [MASK] 虽 [MASK] 以 议 论 入 诗 ， 但 对 仗 工 整 ， [MASK] 言 [MASK] 度 概 括 ， 而 且 [MASK] [MASK] 肯 [MASK] ， 含 蕴 丰 富 ， 很 有 哲 理 性 。 [SEP]
        # masked_lm_positions tokens中每个词在vocabt.txt 中位置
        # masked_lm_labels 是每个MASK掉的词
        (tokens, masked_lm_positions,
         masked_lm_labels) = create_masked_lm_predictions(
             tokens, masked_lm_prob, max_predictions_per_seq, vocab_words, rng)
        instance = TrainingInstance(
            tokens=tokens,
            segment_ids=segment_ids,
            is_random_next=is_random_next,
            masked_lm_positions=masked_lm_positions,
            masked_lm_labels=masked_lm_labels)
        instances.append(instance)
      current_chunk = []
      current_length = 0
    i += 1

  return instances

3.5创建masked LM的预测 

def create_masked_lm_predictions(tokens, masked_lm_prob,
                                 max_predictions_per_seq, vocab_words, rng):
  """Creates the predictions for the masked LM objective."""

  cand_indexes = []
  for (i, token) in enumerate(tokens):
    if token == "[CLS]" or token == "[SEP]":
      continue
    cand_indexes.append(i)

  rng.shuffle(cand_indexes)

  output_tokens = list(tokens)

  num_to_predict = min(max_predictions_per_seq,
                       max(1, int(round(len(tokens) * masked_lm_prob))))

  masked_lms = []
  covered_indexes = set()
  for index in cand_indexes:
    if len(masked_lms) >= num_to_predict:
      break
    if index in covered_indexes:
      continue
    covered_indexes.add(index)

    masked_token = None
    # 80% of the time, replace with [MASK]
    if rng.random() < 0.8:
      masked_token = "[MASK]"
    else:
      # 10% of the time, keep original
      if rng.random() < 0.5:
        masked_token = tokens[index]
      # 10% of the time, replace with random word
      else:
        masked_token = vocab_words[rng.randint(0, len(vocab_words) - 1)]

    output_tokens[index] = masked_token

    masked_lms.append(MaskedLmInstance(index=index, label=tokens[index]))

  masked_lms = sorted(masked_lms, key=lambda x: x.index)

  masked_lm_positions = []
  masked_lm_labels = []
  for p in masked_lms:
    masked_lm_positions.append(p.index)
    masked_lm_labels.append(p.label)
  # output_tokens, mask后的序列， masked_lm_positions，masked的位置
  # masked_lm_labels masked后的真实字
  return (output_tokens, masked_lm_positions, masked_lm_labels)

 选择字符用[MASK]进行替换，调整数据格式，返回。

 3.6截断序列长度

# 将一对序列截断到最大序列长度 
def truncate_seq_pair(tokens_a, tokens_b, max_num_tokens, rng):
  """Truncates a pair of sequences to a maximum sequence length."""
  while True:
    total_length = len(tokens_a) + len(tokens_b)
    if total_length <= max_num_tokens:
      break

    trunc_tokens = tokens_a if len(tokens_a) > len(tokens_b) else tokens_b
    assert len(trunc_tokens) >= 1

    # We want to sometimes truncate from the front and sometimes from the
    # back to add more randomness and avoid biases.
    if rng.random() < 0.5:
      del trunc_tokens[0]
    else:
      trunc_tokens.pop()

4 main()方法

def main(_):
  tf.logging.set_verbosity(tf.logging.INFO)

  tokenizer = tokenization.FullTokenizer(
      vocab_file=FLAGS.vocab_file, do_lower_case=FLAGS.do_lower_case)

  input_files = []  # 输入文件列表
  for input_pattern in FLAGS.input_file.split(","):
    input_files.extend(tf.gfile.Glob(input_pattern))

  tf.logging.info("*** Reading from input files ***")
  for input_file in input_files:
    tf.logging.info("  %s", input_file)

  rng = random.Random(FLAGS.random_seed)
  instances = create_training_instances(
      input_files, tokenizer, FLAGS.max_seq_length, FLAGS.dupe_factor,
      FLAGS.short_seq_prob, FLAGS.masked_lm_prob, FLAGS.max_predictions_per_seq,
      rng)

  output_files = FLAGS.output_file.split(",")
  tf.logging.info("*** Writing to output files ***")
  for output_file in output_files:
    tf.logging.info("  %s", output_file)

  write_instance_to_example_files(instances, tokenizer, FLAGS.max_seq_length,
                                  FLAGS.max_predictions_per_seq, output_files)

加载输入数据，处理成制定格式，存入二进制输出文件中。

主函数入口

if __name__ == "__main__":
  flags.mark_flag_as_required("input_file")
  flags.mark_flag_as_required("output_file")
  flags.mark_flag_as_required("vocab_file")
  tf.app.run()

定义必须的函数并运行。