【动手学习pytorch笔记】36.Transformer实现

Transformer实现

把前面几节的内容拼起来，muti-head Attention、positional encoding…

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

名字挺高大上，实际就是一个单隐藏层的MLP

#@save
class PositionWiseFFN(nn.Module):
    """基于位置的前馈网络"""
    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)))

pytorch中的linear：当输入维度不是二维时，把前面的维度都当作样本维，最后一个维度当作特征维

也是因为输入不再是二维而是三维的，所以起了个高大上的名字叫基于位置的前馈网络

测试一下：

ffn = PositionWiseFFN(4, 4, 8)
ffn.eval()
ffn(torch.ones((2, 3, 4)))[0]

特征维：4->4->8

输出

tensor([[ 0.3839, -0.1441, -0.0172,  0.0031, -0.2300,  0.3045,  0.1544, -0.0293],
        [ 0.3839, -0.1441, -0.0172,  0.0031, -0.2300,  0.3045,  0.1544, -0.0293],
        [ 0.3839, -0.1441, -0.0172,  0.0031, -0.2300,  0.3045,  0.1544, -0.0293]],
       grad_fn=)

对比不同维度的层归一化（LN）和批量归一化（BN）的区别

ln = nn.LayerNorm(2)
bn = nn.BatchNorm1d(2)
X = torch.tensor([[1, 2], [2, 3]], dtype=torch.float32)
# 在训练模式下计算X的均值和方差
print('layer norm:', ln(X), '\nbatch norm:', bn(X))

输出

layer norm: tensor([[-1.0000,  1.0000],
        [-1.0000,  1.0000]], grad_fn=<NativeLayerNormBackward0>) 
batch norm: tensor([[-1.0000, -1.0000],
        [ 1.0000,  1.0000]], grad_fn=<NativeBatchNormBackward0>)

LN是每个样本自己均值为0，方差为1

BN是多个样本的同一特征，均值为0，方差为1

残差链接和LN

#@save
class AddNorm(nn.Module):
    """残差连接后进行层规范化"""
    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)

输出Y做dropout在和输入X相加

测试一下

add_norm = AddNorm([3, 4], 0.5)
add_norm.eval()
add_norm(torch.ones((2, 3, 4)), torch.ones((2, 3, 4))).shape

输出

torch.Size([2, 3, 4])

编码器的Transformer块

#@save
class EncoderBlock(nn.Module):
    """transformer编码器块"""
    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))

注意力->残差连接、LN->MLP->残差连接、LN

测试

X = torch.ones((2, 100, 24))
valid_lens = torch.tensor([3, 2])
encoder_blk = EncoderBlock(24, 24, 24, 24, [100, 24], 24, 48, 8, 0.5)
encoder_blk.eval()
encoder_blk(X, valid_lens).shape

输出

torch.Size([2, 100, 24])

Transformer块，输入和输出形状不变

Transformer编码器（完整）

#@save
class TransformerEncoder(d2l.Encoder):
    """transformer编码器"""
    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))
        # 方便后面画图，记录一下权重
        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

初始化就是把很多个块连接起来

正向传播X先做词嵌入映射，再求位置编码，再相加

encoder = TransformerEncoder(
    200, 24, 24, 24, 24, [100, 24], 24, 48, 8, 2, 0.5)
encoder.eval()
encoder(torch.ones((2, 100), dtype=torch.long), valid_lens).shape

参数：vocab_size, key_size, query_size, value_size,num_hiddens, norm_shape, ffn_num_input, ffn_num_hiddens,num_heads, num_layers, dropout

torch.Size([2, 100, 24])

解码器的Transformer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

解码器的Transformer块 要更复杂一些，这是我们要实现的

直接看forward函数

enc_outputs, enc_valid_lens = state[0], state[1]

enc_outputs, enc_valid_lens是编码器的输出，保存在state中，state中还有一个state[2][self.i]是在预测时用到的，i是第几个transformer块，state[2]用来保存解码器的输入，即key_values = torch.cat((state[2][self.i], X), axis=1)把当前的输入X和之前的state[2][self.i] concat起来作为key_values然后去做Attention，所以在预测阶段，也不需要enc_valid_lens，因为本身就没有后面的输入，本来就看不见。enc_valid_lens是用在训练阶段的，因为输入都是正确的X，后面是有的，但是不让解码器看，所以告诉一个valid_lens去做带掩码的注意力机制。

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，好好理解这句话，什么意思呢，先是第i个时间步，也就是说一个时间步，经过n个解码器块，最终得到输出，进入下一个时间步，得到一个新的X，每一层的state[2][self.i]再去进行拼接，并不是一层一层的拼接，而是每一层独自在没一个时间步的拼接。

测试一下

decoder_blk = DecoderBlock(24, 24, 24, 24, [100, 24], 24, 48, 8, 0.5, 0)
decoder_blk.eval()
X = torch.ones((2, 100, 24))
state = [encoder_blk(X, valid_lens), valid_lens, [None]]
decoder_blk(X, state)[0].shape

输出

torch.Size([2, 100, 24])

形状也不会发生变化

Transformer解码器（完整）

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

init_state 中也像之前讲的 return [enc_outputs, enc_valid_lens, [None] * self.num_layers][0] [1] [2]三个状态[2]是预测时用到的

return self.dense(X)最后做一个线性层输出

训练

因为之前的架构比较好，seq2seq的编码器解码器结构，只是换了里面的实现，所以直接就可以拿来训练了

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)

loss 0.031, 4303.5 tokens/sec on cuda:0

做预测

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}')

go . => va !,  bleu 1.000
i lost . => j'ai perdu .,  bleu 1.000
he's calm . => il est calme .,  bleu 1.000
i'm home . => je suis chez moi .,  bleu 1.000

holy shit! 这bleu score，当然也是数据集比较小的原因。

最后画图看看attention权重

enc_attention_weights = torch.cat(net.encoder.attention_weights, 0).reshape((num_layers, num_heads,
    -1, num_steps))
enc_attention_weights.shape

torch.Size([2, 4, 10, 10])

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))

纵轴query，横轴key-value对，每个头q都重点看了哪些kv，上图是编码器的，后面还有解码器的，和编码器-解码器之间的那个Attention，其实都差不多，这里就省略了。