transformer t5代码解读4(主要内容bert4keras实现t5模型)

回到t5的整体结构之中

回到t5模型的整体的结构之中

(0): T5LayerSelfAttention(
            (SelfAttention): T5Attention(
              (q): Linear(in_features=512, out_features=512, bias=False)
              (k): Linear(in_features=512, out_features=512, bias=False)
              (v): Linear(in_features=512, out_features=512, bias=False)
              (o): Linear(in_features=512, out_features=512, bias=False)
              (relative_attention_bias): Embedding(32, 8)
            )
            (layer_norm): T5LayerNorm()
            (dropout): Dropout(p=0.1, inplace=False)
          )
          (1): T5LayerFF(
            (DenseReluDense): T5DenseReluDense(
              (wi): Linear(in_features=512, out_features=2048, bias=False)
              (wo): Linear(in_features=2048, out_features=512, bias=False)
              (dropout): Dropout(p=0.1, inplace=False)
            )
            (layer_norm): T5LayerNorm()
            (dropout): Dropout(p=0.1, inplace=False)
          )

这里我们接着查看一下T5DenseReluDense模型的结构特点

class T5DenseReluDense(nn.Module):
    def __init__(self, config):
        super().__init__()
        self.wi = nn.Linear(config.d_model, config.d_ff, bias=False)
        self.wo = nn.Linear(config.d_ff, config.d_model, bias=False)
        self.dropout = nn.Dropout(config.dropout_rate)

    def forward(self, hidden_states):
        hidden_states = self.wi(hidden_states)
        hidden_states = nn.functional.relu(hidden_states)
        hidden_states = self.dropout(hidden_states)
        hidden_states = self.wo(hidden_states)
        return hidden_states

这里内容的参数之中，

config.d_model = 512
config.d_ff = 2048

这里还需要注意一下T5LayerNorm网络层结构的实现
先copy一下T5LayerNorm整个部分的源代码

class T5LayerNorm(nn.Module):
    def __init__(self, hidden_size, eps=1e-6):
        """
        Construct a layernorm module in the T5 style No bias and no subtraction of mean.
        """
        super().__init__()
        self.weight = nn.Parameter(torch.ones(hidden_size))
        self.variance_epsilon = eps

    def forward(self, hidden_states):
        # layer norm should always be calculated in float32
        variance = hidden_states.to(torch.float32).pow(2).mean(-1, keepdim=True)
        hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)

        # convert into float16 if necessary
        if self.weight.dtype == torch.float16:
            hidden_states = hidden_states.to(torch.float16)
        return self.weight * hidden_states

常规的layernormalization的计算公式为
$y=\frac{x-\mu }{\sqrt{Var(x)+ϵ}}$
这里的$\mu$为期望，$ϵ$为方差
而新更新的layernormalization的计算公式为
$y=weight\ast$