bert模型源码详细解读

一.bert配置参数解读 bert_config.json

{
  "attention_probs_dropout_prob": 0.1, #乘法attention时，softmax后dropout概率 
  "directionality": "bidi", 
  "hidden_act": "gelu", # 激活函数
  "hidden_dropout_prob": 0.1, #隐藏层dropout概率
  "hidden_size": 768, # 词向量的维度
  "initializer_range": 0.02, # 初始化范围
  "intermediate_size": 3072, # 升维维度
  "max_position_embeddings": 512, # 最大的
  "num_attention_heads": 12, # 总的头数
  "num_hidden_layers": 12, #隐藏层数 ，也就是transformer的encode运行的次数
  "pooler_fc_size": 768, 
  "pooler_num_attention_heads": 12, 
  "pooler_num_fc_layers": 3, 
  "pooler_size_per_head": 128, 
  "pooler_type": "first_token_transform", 
  "type_vocab_size": 2, #segment_ids类别 [0,1]
  "vocab_size": 21128 #词典中词数
}

二. bert模型的输入：

1.config ： bert预训练模型的配置文件路径
2. input_ids ：[batch_size,max_seq_length]
3. input_mask: 是否有掩码 [batch_size, seq_length] 全是1
4. token_type_ids：是否有上下文信息 [batch_size, seq_length] 全是0

class BertModel(object):
  def __init__(self,
               config,
               is_training,
               input_ids,
               input_mask=None,
               token_type_ids=None,
               use_one_hot_embeddings=False,
               scope=None):
    config = copy.deepcopy(config)
    if not is_training:
      config.hidden_dropout_prob = 0.0
      config.attention_probs_dropout_prob = 0.0

    input_shape = get_shape_list(input_ids, expected_rank=2)
    batch_size = input_shape[0]
    seq_length = input_shape[1]

    if input_mask is None:
      input_mask = tf.ones(shape=[batch_size, seq_length], dtype=tf.int32)

    if token_type_ids is None:
      token_type_ids = tf.zeros(shape=[batch_size, seq_length], dtype=tf.int32)

    with tf.variable_scope(scope, default_name="bert"):
      with tf.variable_scope("embeddings"):
        # Perform embedding lookup on the word ids.
        # embedding_output = [batch_size, max_seq_length, embedding_size].
        # embedding_table = [vocab_size, embedding_size]
        (self.embedding_output, self.embedding_table) = embedding_lookup(
            input_ids=input_ids,
            vocab_size=config.vocab_size,
            embedding_size=config.hidden_size,
            initializer_range=config.initializer_range,
            word_embedding_name="word_embeddings",
            use_one_hot_embeddings=use_one_hot_embeddings)

        # Add positional embeddings and token type embeddings, then layer
        # normalize and perform dropout.
        # 将矩阵加入位置信息，再经过nl和dropout
        # 位置信息就是一个大矩阵，是需要训练的，并不是transformer里面那样加的
        self.embedding_output = embedding_postprocessor(
            input_tensor=self.embedding_output,
            use_token_type=True,
            token_type_ids=token_type_ids,
            token_type_vocab_size=config.type_vocab_size,
            token_type_embedding_name="token_type_embeddings",
            use_position_embeddings=True,
            position_embedding_name="position_embeddings",
            initializer_range=config.initializer_range,
            max_position_embeddings=config.max_position_embeddings,
            dropout_prob=config.hidden_dropout_prob)

      with tf.variable_scope("encoder"):
        # This converts a 2D mask of shape [batch_size, seq_length] to a 3D
        # mask of shape [batch_size, seq_length, seq_length] which is used
        # for the attention scores.
        # [batch_size, seq_length, seq_length]
        attention_mask = create_attention_mask_from_input_mask(
            input_ids, input_mask)

        # Run the stacked transformer.
        # `sequence_output` shape = [batch_size, seq_length, hidden_size].
        self.all_encoder_layers = transformer_model(
            input_tensor=self.embedding_output,
            attention_mask=attention_mask,
            hidden_size=config.hidden_size,
            num_hidden_layers=config.num_hidden_layers,
            num_attention_heads=config.num_attention_heads,
            intermediate_size=config.intermediate_size,
            intermediate_act_fn=get_activation(config.hidden_act),
            hidden_dropout_prob=config.hidden_dropout_prob,
            attention_probs_dropout_prob=config.attention_probs_dropout_prob,
            initializer_range=config.initializer_range,
            do_return_all_layers=True)

      self.sequence_output = self.all_encoder_layers[-1]
      # The "pooler" converts the encoded sequence tensor of shape
      # [batch_size, seq_length, hidden_size] to a tensor of shape
      # [batch_size, hidden_size]. This is necessary for segment-level
      # (or segment-pair-level) classification tasks where we need a fixed
      # dimensional representation of the segment.
      with tf.variable_scope("pooler"):
        # We "pool" the model by simply taking the hidden state corresponding
        # to the first token. We assume that this has been pre-trained
        first_token_tensor = tf.squeeze(self.sequence_output[:, 0:1, :], axis=1)
        self.pooled_output = tf.layers.dense(
            first_token_tensor,
            config.hidden_size,
            activation=tf.tanh,
            kernel_initializer=create_initializer(config.initializer_range))

三.embedding_lookup层

得到的输出shape是 ：

embedding_output = [batch_size,max_seq_length, embedding_size].

embedding_table = [vocab_size, embedding_size]

(self.embedding_output, self.embedding_table) = embedding_lookup(
            input_ids=input_ids,
            vocab_size=config.vocab_size,
            embedding_size=config.hidden_size,
            initializer_range=config.initializer_range,
            word_embedding_name="word_embeddings",
            use_one_hot_embeddings=use_one_hot_embeddings)
            
def embedding_look