1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186 | import torch.nn as nn
import torch
import math
class PositionalEncoding(nn.Module):
def __init__(self, d_model, dropout=0.1, max_len=5000):
super(PositionalEncoding, self).__init__()
self.dropout = nn.Dropout(p=dropout)
pe = torch.zeros(max_len, d_model)
position = torch.arange(0, max_len).unsqueeze(1)
_2i = torch.arange(0, d_model, 2)
item = 1 / (10000 ** (_2i / d_model))
pe[:, 0::2] = torch.sin(position * item)
pe[:, 1::2] = torch.cos(position * item)
pe = pe.unsqueeze(0)
self.register_buffer('pe', pe)
def forward(self, x):
x = x + self.pe[:, :x.size(1)]
return self.dropout(x)
class MutiHeadSelfAttention(nn.Module):
def __init__(self, d_model, n_head, dropout=0.1):
super(MutiHeadSelfAttention, self).__init__()
assert d_model % n_head == 0
self.d_k = d_model // n_head
self.n_head = n_head
self.w_q = nn.Linear(d_model, d_model)
self.w_k = nn.Linear(d_model, d_model)
self.w_v = nn.Linear(d_model, d_model)
self.w_o = nn.Linear(d_model, d_model)
self.dropout = nn.Dropout(p=dropout)
def forward(self, q, k, v, mask=None):
batch_size = q.size(0)
q = self.w_q(q).view(batch_size, -1, self.n_head, self.d_k).transpose(1, 2)
k = self.w_k(k).view(batch_size, -1, self.n_head, self.d_k).transpose(1, 2)
v = self.w_v(v).view(batch_size, -1, self.n_head, self.d_k).transpose(1, 2)
output, attention = self.attention(q, k, v, mask)
output = output.transpose(1, 2).contiguous().view(batch_size, -1, self.n_head * self.d_k)
return self.w_o(output)
def attention(self, q, k, v, mask=None):
scores = torch.matmul(q, k.transpose(-2, -1)) / math.sqrt(self.d_k)
if mask is not None:
scores = scores.masked_fill(mask == 0, -1e9)
p_attn = nn.functional.softmax(scores, dim=-1)
p_attn = self.dropout(p_attn)
return torch.matmul(p_attn, v), p_attn
class TransformerEncoderLayer(nn.Module):
def __init__(self, d_model, n_head, d_ff, dropout=0.1):
super(TransformerEncoderLayer, self).__init__()
self.self_attn = MutiHeadSelfAttention(d_model, n_head, dropout)
self.pe = PositionalEncoding(d_model, dropout)
self.ffn = nn.Sequential(
nn.Linear(d_model, d_ff),
nn.ReLU(),
nn.Linear(d_ff, d_model)
)
self.norm1 = nn.LayerNorm(d_model)
self.norm2 = nn.LayerNorm(d_model)
self.dropout = nn.Dropout(dropout)
def forward(self, x, mask=None):
x = x + self.pe(x)
x = x + self.dropout(self.self_attn(x, x, x, mask))
x = self.norm1(x)
x = x + self.dropout(self.ffn(x))
x = self.norm2(x)
return x
class TransformerEncoder(nn.Module):
def __init__(self, n_layer, d_model, n_head, d_ff, dropout=0.1):
super(TransformerEncoder, self).__init__()
self.layers = nn.ModuleList([TransformerEncoderLayer(d_model, n_head, d_ff, dropout) for _ in range(n_layer)])
def forward(self, x, mask=None):
for layer in self.layers:
x = layer(x, mask)
return x
class TransformerDecoderLayer(nn.Module):
def __init__(self, d_model, n_head, d_ff, dropout=0.1):
super(TransformerDecoderLayer, self).__init__()
self.self_attn = MutiHeadSelfAttention(d_model, n_head, dropout)
self.src_attn = MutiHeadSelfAttention(d_model, n_head, dropout)
self.pe = PositionalEncoding(d_model, dropout)
self.ffn = nn.Sequential(
nn.Linear(d_model, d_ff),
nn.ReLU(),
nn.Linear(d_ff, d_model)
)
self.norm1 = nn.LayerNorm(d_model)
self.norm2 = nn.LayerNorm(d_model)
self.norm3 = nn.LayerNorm(d_model)
self.dropout = nn.Dropout(dropout)
def forward(self, x, memory, src_mask=None, tgt_mask=None):
x = x + self.pe(x)
x = x + self.dropout(self.self_attn(x, x, x, tgt_mask))
x = self.norm1(x)
x = x + self.dropout(self.src_attn(x, memory, memory, src_mask))
x = self.norm2(x)
x = x + self.dropout(self.ffn(x))
x = self.norm3(x)
return x
class TransformerDecoder(nn.Module):
def __init__(self, n_layer, d_model, n_head, d_ff, dropout=0.1):
super(TransformerDecoder, self).__init__()
self.layers = nn.ModuleList([TransformerDecoderLayer(d_model, n_head, d_ff, dropout) for _ in range(n_layer)])
def forward(self, x, memory, src_mask=None, tgt_mask=None):
for layer in self.layers:
x = layer(x, memory, src_mask, tgt_mask)
return x
class Transformer(nn.Module):
def __init__(self, n_layer, d_model, n_head, d_ff, dropout=0.1):
super(Transformer, self).__init__()
self.encoder = TransformerEncoder(n_layer, d_model, n_head, d_ff, dropout)
self.decoder = TransformerDecoder(n_layer, d_model, n_head, d_ff, dropout)
def forward(self, src, tgt, src_mask=None, tgt_mask=None):
memory = self.encoder(src, src_mask)
return self.decoder(tgt, memory, src_mask, tgt_mask)
if __name__ == '__main__':
# 参数设置
n_layer = 6
d_model = 256
n_head = 8
d_ff = 2048
dropout = 0.1
src_len = 10
tgt_len = 20
batch_size = 32
# 创建 Transformer 实例
model = Transformer(n_layer, d_model, n_head, d_ff, dropout)
# 生成随机输入数据
# 示例输入数据
# src = torch.tensor([
# [[1, 2, 3], [4, 5, 6], [0, 0, 0], [7, 8, 9]],
# [[0, 0, 0], [1, 1, 1], [2, 2, 2], [0, 0, 0]],
# [[1, 2, 3], [4, 5, 6], [0, 0, 0], [7, 8, 9]],
# [[0, 0, 0], [1, 1, 1], [2, 2, 2], [0, 0, 0]],
# ])
src = torch.randn(batch_size, src_len, d_model)
tgt = torch.randn(batch_size, tgt_len, d_model)
# 生成填充掩码
src_pad_mask = (src != 0).all(dim=-1).unsqueeze(1).unsqueeze(2) # (batch_size, 1, 1, src_len)
# tensor([[[[ True, True, False, True]]],
# [[[False, True, True, False]]],
# [[[ True, True, False, True]]],
# [[[False, True, True, False]]]])
tgt_pad_mask = (tgt != 0).all(dim=-1).unsqueeze(1).unsqueeze(2) # (batch_size, 1, 1, tgt_len)
# 生成未来掩码
future_mask = torch.tril(torch.ones((tgt_len, tgt_len)), diagonal=0).bool()
# tensor([[ True, False, False, False, False],
# [ True, True, False, False, False],
# [ True, True, True, False, False],
# [ True, True, True, True, False],
# [ True, True, True, True, True]])
# 合并掩码
future_mask = future_mask.unsqueeze(0).unsqueeze(1) # (1, 1, tgt_len, tgt_len)
tgt_mask = tgt_pad_mask & future_mask # (batch_size, 1, tgt_len, tgt_len)
# 前向传播
output = model(src, tgt, src_pad_mask, tgt_mask)
print(output.size())
# 检查输出形状
assert output.shape == (batch_size, tgt_len, d_model), f"Expected output shape {(batch_size, tgt_len, d_model)}, but got {output.shape}"
print("Test passed!")
|