Transformer

（1）整体架构

 编码器和解码器都有n层transformer块，相对应的是seq2seq中的RNN。编码器最后一层的输出作为解码器的输入。

（2）实现细节

多头注意力：自注意力kqv（通道数（隐藏层）分组）希望模型提取不同的特征，在输入注意力前，对kqv分组生成不同的权重，即加入全连接层。最后每一组kqv加入注意力，最后在head维度拼接。输入 `X` 的形状: (`batch_size`, 查询或者“键－值”对的个数, `num_hiddens`). 输出 `X` 的形状: (`batch_size`, 查询或者“键－值”对的个数, `num_heads`,  `num_hiddens` / `num_heads`)

 数学形式

在解码的时候，在预测时不应该考虑后边的信息，解决方法就是加入有效数据列长度。

基于位置的前馈神经网络：打平方式不同，因为n是不同的

 add和norm模块，add就是残差连接，norm用layernorm归一化，因为N不统一，所以每个batch归一化。

 信息传递

 预测

（3）总结

（4）多头注意力实现

import math
import pandas as pd
import torch
from torch import nn
from d2l import torch as d2l

#多头注意力
class MultiHeadAttention(nn.Module):
    def __init__(self, key_size, query_size, value_size, num_hiddens,
                 num_heads, dropout, bias=False, **kwargs):
        super(MultiHeadAttention, self).__init__(**kwargs)
        self.num_heads = num_heads
        self.attention = d2l.DotProductAttention(dropout)#选择缩放点积注意力
        self.W_q = nn.Linear(query_size, num_hiddens, bias=bias)
        self.W_k = nn.Linear(key_size, num_hiddens, bias=bias)
        self.W_v = nn.Linear(value_size, num_hiddens, bias=bias)
        self.W_o = nn.Linear(num_hiddens, num_hiddens, bias=bias)

    def forward(self, queries, keys, values, valid_lens):
        # `queries`, `keys`, or `values` 的形状:
        # (`batch_size`, 查询或者“键－值”对的个数, `num_hiddens`)
        # `valid_lens`　的形状:
        # (`batch_size`,) or (`batch_size`, 查询的个数)
        # 经过变换后，输出的 `queries`, `keys`, or `values`　的形状:
        # (`batch_size` * `num_heads`, 查询或者“键－值”对的个数,
        # `num_hiddens` / `num_heads`)
        queries = transpose_qkv(self.W_q(queries), self.num_heads)
        keys = transpose_qkv(self.W_k(keys), self.num_heads)
        values = transpose_qkv(self.W_v(values), self.num_heads)

        if valid_lens is not None:
            # 在轴 0，将第一项（标量或者矢量）复制 `num_heads` 次，
            # 然后如此复制第二项，然后诸如此类。
            valid_lens = torch.repeat_interleave(
                valid_lens, repeats=self.num_heads, dim=0)

        # `output` 的形状: (`batch_size` * `num_heads`, 查询的个数,
        # `num_hiddens` / `num_heads`)
        output = self.attention(queries, keys, values, valid_lens)

        # `output_concat` 的形状: (`batch_size`, 查询的个数, `num_hiddens`)
        output_concat = transpose_output(output, self.num_heads)
        return self.W_o(output_concat)

#使多头并行计算
def transpose_qkv(X, num_heads):
    # 输入 `X` 的形状: (`batch_size`, 查询或者“键－值”对的个数, `num_hiddens`).
    # 输出 `X` 的形状: (`batch_size`, 查询或者“键－值”对的个数, `num_heads`,
    # `num_hiddens` / `num_heads`)
    X = X.reshape(X.shape[0], X.shape[1], num_heads, -1)

    # 输出 `X` 的形状: (`batch_size`, `num_heads`, 查询或者“键－值”对的个数,
    # `num_hiddens` / `num_heads`)
    X = X.permute(0, 2, 1, 3)

    # `output` 的形状: (`batch_size` * `num_heads`, 查询或者“键－值”对的个数,
    # `num_hiddens` / `num_heads`)
    return X.reshape(-1, X.shape[2], X.shape[3])


#
def transpose_output(X, num_heads):
    """逆转 `transpose_qkv` 函数的操作"""
    X = X.reshape(-1, num_heads, X.shape[1], X.shape[2])
    X = X.permute(0, 2, 1, 3)
    return X.reshape(X.shape[0], X.shape[1], -1)

 (5)transformer实现

import math
import pandas as pd
import torch
from torch import nn
from d2l import torch as d2l

#基于位置的前馈网络
class PositionWiseFFN(nn.Module):
    def __init__(self, ffn_num_input, ffn_num_hiddens, ffn_num_outputs,
                 **kwargs):
        super(PositionWiseFFN, self).__init__(**kwargs)
        self.dense1 = nn.Linear(ffn_num_input, ffn_num_hiddens)
        self.relu = nn.ReLU()
        self.dense2 = nn.Linear(ffn_num_hiddens, ffn_num_outputs)

    def forward(self, X):
        return self.dense2(self.relu(self.dense1(X)))#自动将前边维度合并

#下面的例子显示，改变张量的最里层维度的尺寸，会改变成基于位置的前馈网络的输出尺寸。
ffn = PositionWiseFFN(4, 4, 8)
ffn.eval()
ffn(torch.ones((2, 3, 4)))[0]

#残差连接和层归一化
class AddNorm(nn.Module):
    def __init__(self, normalized_shape, dropout, **kwargs):
        super(AddNorm, self).__init__(**kwargs)
        self.dropout = nn.Dropout(dropout)
        self.ln = nn.LayerNorm(normalized_shape)

    def forward(self, X, Y):
        return self.ln(self.dropout(Y) + X)

#多头自注意力和基于位置的前馈网络，这两个子层都使用了残差连接和紧随的层归一化。
class EncoderBlock(nn.Module):
    def __init__(self, key_size, query_size, value_size, num_hiddens,
                 norm_shape, ffn_num_input, ffn_num_hiddens, num_heads,
                 dropout, use_bias=False, **kwargs):
        super(EncoderBlock, self).__init__(**kwargs)
        self.attention = d2l.MultiHeadAttention(
            key_size, query_size, value_size, num_hiddens, num_heads, dropout,
            use_bias)
        self.addnorm1 = AddNorm(norm_shape, dropout)
        self.ffn = PositionWiseFFN(
            ffn_num_input, ffn_num_hiddens, num_hiddens)
        self.addnorm2 = AddNorm(norm_shape, dropout)

    def forward(self, X, valid_lens):
        Y = self.addnorm1(X, self.attention(X, X, X, valid_lens))
        return self.addnorm2(Y, self.ffn(Y))

#堆叠了 num_layers 个 EncoderBlock 类的实例
class TransformerEncoder(d2l.Encoder):
    def __init__(self, vocab_size, key_size, query_size, value_size,
                 num_hiddens, norm_shape, ffn_num_input, ffn_num_hiddens,
                 num_heads, num_layers, dropout, use_bias=False, **kwargs):
        super(TransformerEncoder, self).__init__(**kwargs)
        self.num_hiddens = num_hiddens
        self.embedding = nn.Embedding(vocab_size, num_hiddens)
        self.pos_encoding = d2l.PositionalEncoding(num_hiddens, dropout)
        self.blks = nn.Sequential()
        for i in range(num_layers):
            self.blks.add_module("block"+str(i),
                EncoderBlock(key_size, query_size, value_size, num_hiddens,
                             norm_shape, ffn_num_input, ffn_num_hiddens,
                             num_heads, dropout, use_bias))

    def forward(self, X, valid_lens, *args):
        # 因为位置编码值在 -1 和 1 之间，
        # 因此嵌入值乘以嵌入维度的平方根进行缩放，
        # 然后再与位置编码相加。
        X = self.pos_encoding(self.embedding(X) * math.sqrt(self.num_hiddens))
        self.attention_weights = [None] * len(self.blks)
        for i, blk in enumerate(self.blks):
            X = blk(X, valid_lens)
            self.attention_weights[
                i] = blk.attention.attention.attention_weights
        return X

#解码器块
class DecoderBlock(nn.Module):
    """解码器中第 i 个块"""
    def __init__(self, key_size, query_size, value_size, num_hiddens,
                 norm_shape, ffn_num_input, ffn_num_hiddens, num_heads,
                 dropout, i, **kwargs):
        super(DecoderBlock, self).__init__(**kwargs)
        self.i = i
        self.attention1 = d2l.MultiHeadAttention(
            key_size, query_size, value_size, num_hiddens, num_heads, dropout)
        self.addnorm1 = AddNorm(norm_shape, dropout)
        self.attention2 = d2l.MultiHeadAttention(
            key_size, query_size, value_size, num_hiddens, num_heads, dropout)
        self.addnorm2 = AddNorm(norm_shape, dropout)
        self.ffn = PositionWiseFFN(ffn_num_input, ffn_num_hiddens,
                                   num_hiddens)
        self.addnorm3 = AddNorm(norm_shape, dropout)

    def forward(self, X, state):
        enc_outputs, enc_valid_lens = state[0], state[1]
        # 训练阶段，输出序列的所有词元都在同一时间处理，
        # 因此 `state[2][self.i]` 初始化为 `None`。
        # 预测阶段，输出序列是通过词元一个接着一个解码的，
        # 因此 `state[2][self.i]` 包含着直到当前时间步第 `i` 个块解码的输出表示
        if state[2][self.i] is None:
            key_values = X
        else:
            key_values = torch.cat((state[2][self.i], X), axis=1)
        state[2][self.i] = key_values
        if self.training:
            batch_size, num_steps, _ = X.shape
            # `dec_valid_lens` 的开头: (`batch_size`, `num_steps`),
            # 其中每一行是 [1, 2, ..., `num_steps`]
            dec_valid_lens = torch.arange(
                1, num_steps + 1, device=X.device).repeat(batch_size, 1)
        else:
            dec_valid_lens = None

        # 自注意力
        X2 = self.attention1(X, key_values, key_values, dec_valid_lens)
        Y = self.addnorm1(X, X2)
        # 编码器－解码器注意力。
        # `enc_outputs` 的开头: (`batch_size`, `num_steps`, `num_hiddens`)
        Y2 = self.attention2(Y, enc_outputs, enc_outputs, enc_valid_lens)
        Z = self.addnorm2(Y, Y2)
        return self.addnorm3(Z, self.ffn(Z)), state

#解码器
class TransformerDecoder(d2l.AttentionDecoder):
    def __init__(self, vocab_size, key_size, query_size, value_size,
                 num_hiddens, norm_shape, ffn_num_input, ffn_num_hiddens,
                 num_heads, num_layers, dropout, **kwargs):
        super(TransformerDecoder, self).__init__(**kwargs)
        self.num_hiddens = num_hiddens
        self.num_layers = num_layers
        self.embedding = nn.Embedding(vocab_size, num_hiddens)
        self.pos_encoding = d2l.PositionalEncoding(num_hiddens, dropout)
        self.blks = nn.Sequential()
        for i in range(num_layers):
            self.blks.add_module("block"+str(i),
                DecoderBlock(key_size, query_size, value_size, num_hiddens,
                             norm_shape, ffn_num_input, ffn_num_hiddens,
                             num_heads, dropout, i))
        self.dense = nn.Linear(num_hiddens, vocab_size)

    def init_state(self, enc_outputs, enc_valid_lens, *args):
        return [enc_outputs, enc_valid_lens, [None] * self.num_layers]

    def forward(self, X, state):
        X = self.pos_encoding(self.embedding(X) * math.sqrt(self.num_hiddens))
        self._attention_weights = [[None] * len(self.blks) for _ in range (2)]
        for i, blk in enumerate(self.blks):
            X, state = blk(X, state)
            # 解码器自注意力权重
            self._attention_weights[0][
                i] = blk.attention1.attention.attention_weights
            # “编码器－解码器”自注意力权重
            self._attention_weights[1][
                i] = blk.attention2.attention.attention_weights
        return self.dense(X), state

    @property
    def attention_weights(self):
        return self._attention_weights

#训练
num_hiddens, num_layers, dropout, batch_size, num_steps = 32, 2, 0.1, 64, 10
lr, num_epochs, device = 0.005, 200, d2l.try_gpu()
ffn_num_input, ffn_num_hiddens, num_heads = 32, 64, 4
key_size, query_size, value_size = 32, 32, 32
norm_shape = [32]

train_iter, src_vocab, tgt_vocab = d2l.load_data_nmt(batch_size, num_steps)

encoder = TransformerEncoder(
    len(src_vocab), key_size, query_size, value_size, num_hiddens,
    norm_shape, ffn_num_input, ffn_num_hiddens, num_heads,
    num_layers, dropout)
decoder = TransformerDecoder(
    len(tgt_vocab), key_size, query_size, value_size, num_hiddens,
    norm_shape, ffn_num_input, ffn_num_hiddens, num_heads,
    num_layers, dropout)
net = d2l.EncoderDecoder(encoder, decoder)
d2l.train_seq2seq(net, train_iter, lr, num_epochs, tgt_vocab, device)

#计算BLUE
engs = ['go .', "i lost .", 'he\'s calm .', 'i\'m home .']
fras = ['va !', 'j\'ai perdu .', 'il est calme .', 'je suis chez moi .']
for eng, fra in zip(engs, fras):
    translation, dec_attention_weight_seq = d2l.predict_seq2seq(
        net, eng, src_vocab, tgt_vocab, num_steps, device, True)
    print(f'{eng} => {translation}, ',
          f'bleu {d2l.bleu(translation, fra, k=2):.3f}')



#编码器自注意力权重的形状为 (编码器层数, 注意力头数, num_steps或查询的数目, num_steps 或“键－值”对的数目) 。
enc_attention_weights = torch.cat(net.encoder.attention_weights, 0).reshape((num_layers, num_heads,
    -1, num_steps))
enc_attention_weights.shape

#接下来，将逐行呈现两层多头注意力的权重。每个注意力头都根据查询、键和值的不同的表示子空间来表示不同的注意力。
d2l.show_heatmaps(
    enc_attention_weights.cpu(), xlabel='Key positions',
    ylabel='Query positions', titles=['Head %d' % i for i in range(1, 5)],
    figsize=(7, 3.5))

dec_attention_weights_2d = [head[0].tolist()
                            for step in dec_attention_weight_seq
                            for attn in step for blk in attn for head in blk]
dec_attention_weights_filled = torch.tensor(
    pd.DataFrame(dec_attention_weights_2d).fillna(0.0).values)
dec_attention_weights = dec_attention_weights_filled.reshape((-1, 2, num_layers, num_heads, num_steps))
dec_self_attention_weights, dec_inter_attention_weights = \
    dec_attention_weights.permute(1, 2, 3, 0, 4)
dec_self_attention_weights.shape, dec_inter_attention_weights.shape

d2l.show_heatmaps(
    dec_inter_attention_weights, xlabel='Key positions',
    ylabel='Query positions', titles=['Head %d' % i for i in range(1, 5)],
    figsize=(7, 3.5))