沐神版《动手学深度学习》学习笔记,记录学习过程,详细的内容请大家购买书籍查阅。
b站视频链接
开源教程链接
沐神Transformer论文逐段精读【论文精读】
Transformer
Transformer的论文:Attention Is All You Need!
Transformer的模型架构
编码器和解码器中的一部分是相同的,解码器多出了Masked的多头注意力。因为要有残差连接,所以要保证数据经过MLP后的维度与原来的维度是相同的,所以在模型中将输出维度都设置为512,这个设计影响了后面的BERT、GPT等工作,可调的参数就只有层数N和模型的输出维度。
如果在写文章时,相关部分有用到了别人的东西,最好在文章里讲一下它是什么东西,不能指望别人都知道所有的细节,能够从几句话讲清楚是不错的。
层归一化
Transformer中使用了Layer Norm,为什么我们在这些变长的应用里不使用Batch Norm,考虑一个最简单的二位输入:
LayerNorm是对样本做标准化,BatchNorm对特征做标准化。在序列不等的情况下使用BatchNorm容易使得方差和均值抖动,LayerNorm是每个样本自己做均值和方差,不需要存全局的信息,相对稳定。
缩放点积注意力
带Mask的注意力计算
保证解码器在计算t时刻时,只能看到t-1时刻及以前的信息,具体实现是将对应信息变得很小,导致在计算注意力矩阵时这些值经过softmax后趋近于0。
多头注意力
缩放点积注意力是没有学习参数的,所以先把V、K、Q投影到线性层,在经过多个注意力计算,最后再拼接在一起做一次投影,总的来说效果上比较像卷积的多通道。
Transformer选择8个头,每个头投影到512/8=64维度上。
Transformer中注意力的使用
第三个多头注意力层是将编码器的输出作为Q和K,将解码器的输出作为V进行注意力计算。
前馈层有两层,第一层将维度扩展到2048,第二层再将维度缩放到512。
Attention和RNN的区别
attention就是将整个序列的全局信息抓取出来做一个汇聚,RNN把上一个时刻的信息传入下一个时刻作为输入的一部分,两个模型的不同之处就在于序列信息的传递。
位置编码
attention中不存在时序信息,所以在transformer的输入中加入时序信息(位置编码)。
Transformer架构
基于编码器解码器架构,
一个transformer的块相当于一个RNN层:
多头注意力有点像卷积中的多通道,希望学到不同粒度的特征:
可学习的参数: W i ( q ) W_i^{(q)} Wi(q) 、 W i ( k ) W_i^{(k)} Wi(k)、 W i ( v ) W_i^{(v)} Wi(v)、 W o W_o Wo:
Attention是没有强时序关系的,当前位置的信息可以看到在后面发生的信息,这在编码器中是合适的,但在解码器中不合适,解码的东西不能考虑该元素和后续的元素。
所以在解码器的多头注意力中,采用有掩码的多头注意力,计算
x
i
x_i
xi的输出时,假装当前序列长度为
i
i
i。
等价于两层核窗口为1的一维卷积层:
批量归一化对每个特征/通道里的元素进行归一化,不适合序列长度会变的NLP应用。
层归一化对每个样本中的元素进行归一化。
两种方法的目的都是想要网络训练更稳定,可以做的更深。
编码器中的输出,将其作为解码器中第i个Transformer块中多头注意力的key和value(query来自于解码器的目标序列):
在做训练时可以把数据一筐的丢进来,在预测的时候按照顺序来进行,一个一个往前走,不再是O(n) -> O(n)指transformer的并行度:
总结
动手学
多头注意力
#@save
def transpose_qkv(X, num_heads):
"""为了多注意力头的并行计算而变换形状"""
# 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)
# 最终输出的形状:(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])
#@save
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)
#@save
class MultiHeadAttention(nn.Module):
"""多头注意力"""
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)
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,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,)或(batch_size,查询的个数)
# 经过变换后,输出的queries,keys,values 的形状:
# (batch_size*num_heads,查询或者“键-值”对的个数,num_hiddens/num_heads)
# 这样做为的是multi-head计算一次attention,而不是每个head都做各自的attention
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) # 逆转回来
return self.W_o(output_concat)
num_hiddens, num_heads = 100, 5
attention = MultiHeadAttention(num_hiddens, num_hiddens, num_hiddens, num_hiddens, num_heads, 0.5)
attention.eval()
MultiHeadAttention(
(attention): DotProductAttention(
(dropout): Dropout(p=0.5, inplace=False)
)
(W_q): Linear(in_features=100, out_features=100, bias=False)
(W_k): Linear(in_features=100, out_features=100, bias=False)
(W_v): Linear(in_features=100, out_features=100, bias=False)
(W_o): Linear(in_features=100, out_features=100, bias=False)
)
batch_size, num_queries = 2, 4
num_kvpairs, valid_lens = 6, torch.tensor([3, 2])
X = torch.ones((batch_size, num_queries, num_hiddens))
Y = torch.ones((batch_size, num_kvpairs, num_hiddens))
attention(X, Y, Y, valid_lens).shape
torch.Size([2, 4, 100])
Transformer
#@save
class PositionWiseFFN(nn.Module):
"""基于位置的前馈网络"""
# 类似一个单隐藏层的MLP,但是输入是三维
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)))
ffn = PositionWiseFFN(4, 4, 8)
ffn.eval()
ffn(torch.ones((2, 3, 4))).shape
torch.Size([2, 3, 8])
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('X',X)
print('layer norm:', ln(X), '\nbatch norm:', bn(X))
X tensor([[1., 2.],
[2., 3.]])
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>)
#@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):
# 先残差再归一化
return self.ln(self.dropout(Y) + X)
add_norm = AddNorm([4], 0.5)
add_norm.eval()
add_norm(torch.ones((2,4)), torch.ones((2,4)))
tensor([[0., 0., 0., 0.],
[0., 0., 0., 0.]], grad_fn=<NativeLayerNormBackward0>)
#@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))
#@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)) # embedding数值偏小,
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
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
torch.Size([2, 100, 24])
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
torch.Size([2, 100, 24])
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
class TransformerDecoder(d2l.AttentionDecoder):
"""Transformer解码器"""
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, **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
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
torch.Size([2, 100, 24])
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)
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 riche ., bleu 0.658
i'm home . => je suis chez moi ., bleu 1.000
观察3个多头注意力的注意力权重矩阵
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))\
# 每一行是一个query,每一列是一个(key, value)pair
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
(torch.Size([2, 4, 6, 10]), torch.Size([2, 4, 6, 10]))
# Plusonetoincludethebeginning-of-sequencetoken
d2l.show_heatmaps(
dec_self_attention_weights[:, :, :, :len(translation.split()) + 1],
xlabel='Key positions', ylabel='Query positions',
titles=['Head %d' % i for i in range(1, 5)], figsize=(7, 3.5))
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))