Transformer结合BERT代码的理解

Self-attention部分：

Batch Norm针对不同样本的同一通道的特征进行归一化（在NLP中表示，不同句子的同一位置的词的特征），要求样本通道的数量基本相同（在NLP中，也就是同一批样本中，句子长度大致一样）。

Layer Norm针对同一样本的不同通道做归一化（在NLP表示，同一个句子的不同词的特征），只针对一个样本。

从上面的区别可以看出，Batch Norm针对整个批次的样本，需要反映出整个批次的分布，所以需要批次的数量比较大。Layer Norm只针对一个样本，与批次的大小无关。

Batch Norm通常应用于CV，Layer Norm通常应用于NLP。

LN是和BN非常近似的一种归一化方法，不同的是BN取的是不同样本的同一个特征，而LN取的是同一个样本的不同特征。在BN和LN都能使用的场景中，BN的效果一般优于LN，原因是基于不同数据，同一特征得到的归一化特征更不容易损失信息。

但是有些场景是不能使用BN的，例如batchsize较小或者在RNN中，这时候可以选择使用LN，LN得到的模型更稳定且起到正则化的作用。RNN能应用到小批量和RNN中是因为LN的归一化统计量的计算是和batchsize没有关系的。

首先是对编码进行embedding。

class BertEmbeddings(nn.Module):
    # 传入句子编码，位置编码，和类型编码，得到最终的嵌入表示
    """Construct the embeddings from word, position and token_type embeddings."""

    def __init__(self, config):
        super().__init__()
        self.word_embeddings = nn.Embedding(config.vocab_size, config.hidden_size, padding_idx=config.pad_token_id)
        self.position_embeddings = nn.Embedding(config.max_position_embeddings, config.hidden_size)
        self.token_type_embeddings = nn.Embedding(config.type_vocab_size, config.hidden_size)
        self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
        self.dropout = nn.Dropout(config.hidden_dropout_prob)

        # position_ids (1, len position emb) is contiguous in memory and exported when serialized
        self.register_buffer("position_ids", torch.arange(config.max_position_embeddings).expand((1, -1)))
        self.position_embedding_type = getattr(config, "position_embedding_type", "absolute")

    def forward(
        self, input_ids=None, token_type_ids=None, position_ids=None, inputs_embeds=None, past_key_values_length=0
    ):
        if input_ids is not None:
            input_shape = input_ids.size()
        else:
            input_shape = inputs_embeds.size()[:-1]

        seq_length = input_shape[1]

        if position_ids is None:
            position_ids = self.position_ids[:, past_key_values_length : seq_length + past_key_values_length]

        if token_type_ids is None:
            token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=self.position_ids.device)

        if inputs_embeds is None:
            inputs_embeds = self.word_embeddings(input_ids)
        token_type_embeddings = self.token_type_embeddings(token_type_ids)

        embeddings = inputs_embeds + token_type_embeddings
        if self.position_embedding_type == "absolute":
            position_embeddings = self.position_embeddings(position_ids)
            embeddings += position_embeddings
        embeddings = self.LayerNorm(embeddings)  # LN可以让模型不容易过拟合
        embeddings = self.dropout(embeddings)  # embedding后添加dropout，防止过拟合，套路
        return embeddings

然后是进行Self-attention的计算(完全按照论文)：

class BertSelfAttention(nn.Module):
    def __init__(self, config):
        super().__init__()
        if config.hidden_size % config.num_attention_heads != 0 and not hasattr(config, "embedding_size"):
            raise ValueError(
                "The hidden size (%d) is not a multiple of the number of attention "
                "heads (%d)" % (config.hidden_size, config.num_attention_heads)
            )

        self.num_attention_heads = config.num_attention_heads
        self.attention_head_size = int(config.hidden_size / config.num_attention_heads)  # 计算每个head的大小
        self.all_head_size = self.num_attention_heads * self.attention_head_size  # hidden_size可能无法整除head的数量

        self.query = nn.Linear(config.hidden_size, self.all_head_size)  # 转换矩阵，将多个attention head合并成一个张量进行计算
        self.key = nn.Linear(config.hidden_size, self.all_head_size)
        self.value = nn.Linear(config.hidden_size, self.all_head_size)

        self.dropout = nn.Dropout(config.attention_probs_dropout_prob)
        self.position_embedding_type = getattr(config, "position_embedding_type", "absolute")
        if self.position_embedding_type == "relative_key" or self.position_embedding_type == "relative_key_query":
            self.max_position_embeddings = config.max_position_embeddings
            self.distance_embedding = nn.Embedding(2 * config.max_position_embeddings - 1, self.attention_head_size)

        self.is_decoder = config.is_decoder

    def transpose_for_scores(self, x):
        new_x_shape = x.size()[:-1] + (self.num_attention_heads, self.attention_head_size)
        x = x.view(*new_x_shape)
        return x.permute(0, 2, 1, 3)

    def forward(
        self,
        hidden_states,
        attention_mask=None,
        head_mask=None,
        encoder_hidden_states=None,
        encoder_attention_mask=None,
        past_key_value=None,
        output_attentions=False,
    ):
        mixed_query_layer = self.query(hidden_states)  # 对输入使用相应的权重矩阵转化为对应的query张量

        # If this is instantiated as a cross-attention module, the keys
        # and values come from an encoder; the attention mask needs to be
        # such that the encoder's padding tokens are not attended to.
        is_cross_attention = encoder_hidden_states is not None  # Transformer的Encoder，Q是来自目标序列

        if is_cross_attention and past_key_value is not None:
            # reuse k,v, cross_attentions
            key_layer = past_key_value[0]
            value_layer = past_key_value[1]
            attention_mask = encoder_attention_mask
        elif is_cross_attention:
            key_layer = self.transpose_for_scores(self.key(encoder_hidden_states))
            value_layer = self.transpose_for_scores(self.value(encoder_hidden_states))
            attention_mask = encoder_attention_mask
        elif past_key_value is not None:
            key_layer = self.transpose_for_scores(self.key(hidden_states))
            value_layer = self.transpose_for_scores(self.value(hidden_states))
            key_layer = torch.cat([past_key_value[0], key_layer], dim=2)
            value_layer = torch.cat([past_key_value[1], value_layer], dim=2)
        else:
            key_layer = self.transpose_for_scores(self.key(hidden_states))
            value_layer = self.transpose_for_scores(self.value(hidden_states))

        query_layer = self.transpose_for_scores(mixed_query_layer)

        if self.is_decoder:
            past_key_value = (key_layer, value_layer)

        # Take the dot product between "query" and "key" to get the raw attention scores.
        attention_scores = torch.matmul(query_layer, key_layer.transpose(-1, -2))

        if attention_mask is not None:
            # Apply the attention mask is (precomputed for all layers in BertModel forward() function)
            attention_scores = attention_scores + attention_mask

        # Normalize the attention scores to probabilities.
        attention_probs = nn.Softmax(dim=-1)(attention_scores)

        # This is actually dropping out entire tokens to attend to, which might
        # seem a bit unusual, but is taken from the original Transformer paper.
        attention_probs = self.dropout(attention_probs)  # 为什么在这里添加dropout

        # Mask heads if we want to
        if head_mask is not None:
            attention_probs = attention_probs * head_mask

        context_layer = torch.matmul(attention_probs, value_layer)  # 得到注意力内容

        context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
        new_context_layer_shape = context_layer.size()[:-2] + (self.all_head_size,)  # 将head的维度还原
        context_layer = context_layer.view(*new_context_layer_shape)

        outputs = (context_layer, attention_probs) if output_attentions else (context_layer,)

        if self.is_decoder:
            outputs = outputs + (past_key_value,)  # gpt2中将过去的key和value追加
        return outputs

Add&Norm操作：

class BertSelfOutput(nn.Module):
    # Add&Norm封装，进行复用（或者在中间添加一些层，比如一个全连接层，另外FFN复用了该类，并且用到了dense层
    def __init__(self, config):
        super().__init__()
        self.dense = nn.Linear(config.hidden_size, config.hidden_size)
        self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
        self.dropout = nn.Dropout(config.hidden_dropout_prob)

    def forward(self, hidden_states, input_tensor):
        hidden_states = self.dense(hidden_states)
        hidden_states = self.dropout(hidden_states)
        # 前面的两部分，在论文中并没有提到，应该实现着自己添加的
        hidden_states = self.LayerNorm(hidden_states + input_tensor)
        return hidden_states

FFN网络：显示用BertIntermediate，然后使用BertSelfOutput

class BertIntermediate(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.dense = nn.Linear(config.hidden_size, config.intermediate_size)
        if isinstance(config.hidden_act, str):
            self.intermediate_act_fn = ACT2FN[config.hidden_act]  # 激活函数字典
        else:
            self.intermediate_act_fn = config.hidden_act

    def forward(self, hidden_states):
        hidden_states = self.dense(hidden_states)
        hidden_states = self.intermediate_act_fn(hidden_states)  # Transformer论文中使用的是relu激活函数
        return hidden_states

至此上面就是一个Bert Layer的全部组件，省略了BertLayer,整合如下：

class BertEncoder(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.config = config
        self.layer = nn.ModuleList([BertLayer(config) for _ in range(config.num_hidden_layers)])  # 模块堆叠

    def forward(
        self,
        hidden_states,
        attention_mask=None,
        head_mask=None,
        encoder_hidden_states=None,
        encoder_attention_mask=None,
        past_key_values=None,
        # use_cache=None,
        output_attentions=False,
        output_hidden_states=False,
        # return_dict=True,
    ):
        all_hidden_states = () if output_hidden_states else None
        all_self_attentions = () if output_attentions else None

        for i, layer_module in enumerate(self.layer):
            if output_hidden_states:
                all_hidden_states = all_hidden_states + (hidden_states,)

            layer_head_mask = head_mask[i] if head_mask is not None else None
            past_key_value = past_key_values[i] if past_key_values is not None else None

            layer_outputs = layer_module(
                hidden_states,
                attention_mask,
                layer_head_mask,
                encoder_hidden_states,
                encoder_attention_mask,
                past_key_value,
                output_attentions,
            )

            hidden_states = layer_outputs[0]  # 更新为最新层的，也就是每层的输出
            # if use_cache:
            #     next_decoder_cache += (layer_outputs[-1],)
            if output_attentions:
                all_self_attentions = all_self_attentions + (layer_outputs[1],)
                if self.config.add_cross_attention:
                    all_cross_attentions = all_cross_attentions + (layer_outputs[2],)

        if output_hidden_states:
            all_hidden_states = all_hidden_states + (hidden_states,)

        
        output = {'last_hidden_state': hidden_states, 'past_key_values': next_decoder_cache,
                  'hidden_states': all_hidden_states, 'attentions': all_self_attentions,
                  'cross_attentions': all_cross_attentions}

        return output

最开始hidden-state是embeddings，后面都是每层的输出，而且维度和embeddings一致，所以可以不断堆叠层次。

不知道这个输出是用来干嘛的，用来作为指针？那为什么不是attention权重作为指针？

class BertPooler(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.dense = nn.Linear(config.hidden_size, config.hidden_size)
        self.activation = nn.Tanh()

    def forward(self, hidden_states):
        # We "pool" the model by simply taking the hidden state corresponding
        # to the first token.
        # 应该是使用每个词的第一个维度进行pool_out,感觉误差会很大，应该是得到指针类似的结果，（-1,1）取值
        first_token_tensor = hidden_states[:, 0]
        pooled_output = self.dense(first_token_tensor)
        pooled_output = self.activation(pooled_output)
        return pooled_output
    

最终的模型：这里作者的写法考虑了通用性。代码是Encoder写法,但是可以通过设置参数的值，当成Decoder使用。

class BertModel(nn.Module):  # 不继承BertPreTrainedModel
    """

    The model can behave as an encoder (with only self-attention) as well as a decoder, in which case a layer of
    cross-attention is added between the self-attention layers, following the architecture described in `Attention is
    all you need <https://arxiv.org/abs/1706.03762>`__ by Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit,
    Llion Jones, Aidan N. Gomez, Lukasz Kaiser and Illia Polosukhin.

    To behave as an decoder the model needs to be initialized with the :obj:`is_decoder` argument of the configuration
    set to :obj:`True`. To be used in a Seq2Seq model, the model needs to initialized with both :obj:`is_decoder`
    argument and :obj:`add_cross_attention` set to :obj:`True`; an :obj:`encoder_hidden_states` is then expected as an
    input to the forward pass.
    """

    def __init__(self, config, add_pooling_layer=True):
        super().__init__()
        self.config = config

        self.embeddings = BertEmbeddings(config)
        self.encoder = BertEncoder(config)

        self.pooler = BertPooler(config) if add_pooling_layer else None

    def get_input_embeddings(self):
        return self.embeddings.word_embeddings

    def set_input_embeddings(self, value):
        self.embeddings.word_embeddings = value

    def forward(
        self,
        input_ids=None,
        attention_mask=None,
        token_type_ids=None,
        position_ids=None,
        head_mask=None,
        inputs_embeds=None,  # 直接传入已经编码号的输入，而不是传入input_ids和position_ids等在进行Embedding
        encoder_hidden_states=None,
        encoder_attention_mask=None,
        past_key_values=None,
        use_cache=None,
        output_attentions=None,
        output_hidden_states=None,
        # return_dict=None,
    ):
        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
        output_hidden_states = (
            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
        )

        if input_ids is not None and inputs_embeds is not None:
            raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
        elif input_ids is not None:
            input_shape = input_ids.size()
            batch_size, seq_length = input_shape
        elif inputs_embeds is not None:
            input_shape = inputs_embeds.size()[:-1]
            batch_size, seq_length = input_shape
        else:
            raise ValueError("You have to specify either input_ids or inputs_embeds")

        device = input_ids.device if input_ids is not None else inputs_embeds.device

        # past_key_values_length
        past_key_values_length = past_key_values[0][0].shape[2] if past_key_values is not None else 0

        if attention_mask is None:
            attention_mask = torch.ones((batch_size, seq_length + past_key_values_length), device=device)
        if token_type_ids is None:
            token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=device)

        # We can provide a self-attention mask of dimensions [batch_size, from_seq_length, to_seq_length]
        # ourselves in which case we just need to make it broadcastable to all heads.
        extended_attention_mask: torch.Tensor = self.get_extended_attention_mask(attention_mask, input_shape, device)

        if self.config.is_decoder and encoder_hidden_states is not None:
            encoder_batch_size, encoder_sequence_length, _ = encoder_hidden_states.size()
            encoder_hidden_shape = (encoder_batch_size, encoder_sequence_length)
            if encoder_attention_mask is None:
                encoder_attention_mask = torch.ones(encoder_hidden_shape, device=device)
            encoder_extended_attention_mask = self.invert_attention_mask(encoder_attention_mask)
        else:
            encoder_extended_attention_mask = None

        # Prepare head mask if needed
        # 1.0 in head_mask indicate we keep the head
        # attention_probs has shape bsz x n_heads x N x N
        # input head_mask has shape [num_heads] or [num_hidden_layers x num_heads]
        # and head_mask is converted to shape [num_hidden_layers x batch x num_heads x seq_length x seq_length]
        head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)

        embedding_output = self.embeddings(
            input_ids=input_ids,
            position_ids=position_ids,
            token_type_ids=token_type_ids,
            inputs_embeds=inputs_embeds,
            past_key_values_length=past_key_values_length,
        )
        encoder_outputs = self.encoder(
            embedding_output,
            attention_mask=extended_attention_mask,
            head_mask=head_mask,
            encoder_hidden_states=encoder_hidden_states,
            encoder_attention_mask=encoder_extended_attention_mask,
            past_key_values=past_key_values,
            # use_cache=use_cache,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            # return_dict=return_dict,
        )
        
        # sequence_output = encoder_outputs[0]
        sequence_output = encoder_outputs['last_hidden_state']
        pooled_output = self.pooler(sequence_output) if self.pooler is not None else None         

        output = {'last_hidden_state': sequence_output, 'pooler_output': pooled_output,
                  'past_key_values': encoder_outputs['past_key_values'], 'hidden_states': encoder_outputs['hidden_states'],
                  'attentions': encoder_outputs['attentions'], 'cross_attentions': encoder_outputs['cross_attentions']
                  }
        
        return output