-
Notifications
You must be signed in to change notification settings - Fork 0
/
Transformer.py
187 lines (149 loc) · 7.08 KB
/
Transformer.py
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
187
import torch
import torch.nn as nn
import math
class InputEmbeddings(nn.Module):
def __init__(self,d_model:int,vocab_size:int):
self.d_model = d_model
self.vocab_size = vocab_size
self.embedding = nn.Embedding(vocab_size,d_model)
def forward(self,x): # !!
return self.embedding(x) * math.sqrt(self.d_model)
class PositionalEncoding(nn.Module):
def __init__(self,d_model:int,seq_len:int,dropout:float)->None:
super().__init__()
self.d_model = d_model
self.seq_len = seq_len
self.dropout = nn.Dropout(dropout)
'''
matriz de forma seq_len,d_model, siendo seq_len tamaño de la frase\n
y d_model tamaño del vector representando a cada palabra.
muchos vectores de longitud 512, el numero\n
de vectores es igual a la cantidad de palabras en la frase
'''
matriz = torch.zeros(seq_len,d_model)
# vector representa la posicion de la palabra en la frase.
pos = torch.arange(0,seq_len,dtype=torch.float).unsqueeze(1)
#!! vector del denominador en positional encoding
div_term = torch.exp(torch.arange(0,d_model,2).float()*(-math.log(10000.0)/d_model))
'''
en la matriz que calculamos, a las posiciones pares\n
se les saca el seno de su posicion multiplicado por el denominador\n
a las posiciones impares se les saca el coseno de su posicion multiplicado por el denominador
'''
matriz[:,0::2] = torch.sin(pos*div_term)
matriz[:,1::2]= torch.cos(pos*div_term)
matriz = matriz.unsqueeze_(0)
# guardar la matriz en el buffer
self.register_buffer('matriz',matriz)
def forward(self,x):
x = x + (self.pos[:,x.shape[1],:]).requires_grad_(False)
return self.dropout(x)
#layer normalization
class LayerNormalization(nn.Module):
def __init__(self,eps:float=10**-6)->None:
super().__init__()
self.eps = eps
self.alpha = nn.Parameter(torch.ones(1))
self.bias = nn.Parameter(torch.zeros(1))
def forward(self,x):
mean = x.mean(dim=-1,keepdim=True)
std = x.std(dim=-1,keepdim=True)
return self.alpha *(x-mean)/(std+self.eps)+self.bias
class FeedForwardBlock(nn.Module):
def __init__(self,d_model:int,d_ff:int,dropout:float) -> None:
super().__init__()
self.linear_1 = nn.Linear(d_model,d_ff)
self.dropout = nn.Dropout(dropout)
self.linear_2 = nn.Linear(d_ff,d_model)
def forward(self,x):
return self.linear_2(self.dropout(torch.relu(self.linear_1(x))))
class MultiHeadAttentionBlock(nn.Module):
def __init__(self, d_model: int, h: int, dropout: float) -> None:
super().__init__()
self.d_model = d_model
self.h = h
assert d_model % h == 0, 'error, not divisible'
self.d_k = d_model // h
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(dropout)
@staticmethod
def attention(query,key,value,mask,dropout:nn.Dropout):
d_k = query.shape[-1]
attention_scores = (query @ key.transpose(-2,-1))/math.sqrt(d_k)
if mask is not None:
attention_scores.masked_fill_(mask == 0,-1e9)
attention_scores = torch.softmax(dim=-1)
if dropout is not None:
attention_scores = dropout(attention_scores)
return (attention_scores @ value),attention_scores
def forward(self,q,k,v,mask):
query = self.W_Q(q)# (Batch,seq_len,d_model) ---> (Batch,seq_len,d_model)
key = self.W_K(k)
value = self.W_V(v)
#(Batch,seq_len,d_model) ---> (Batch,seq_len,h,d_k) ---> (Batch,h,seq_len,d_k)
query = query.view(query.shape[0],query.shape[1],self.h,self.d_k).transpose(1,2)
key = key.view(key.shape[0],key.shape[1],self.h,self.d_k).transpose(1,2)
value = value.view(value.shape[0],value.shape[1],self.h.self.d_k).transpose(1,2)
x,self.attention_scores = MultiHeadAttentionBlock.attention(query,key,value,mask,self.dropout)
x = x.transpose(1,2).contiguous().view(x.shape[0,-1,self.h*self.d_k])
return self.w_o(x)
class ResidualConnection(nn.module):
def __init__(self,dropout:float)->None:
super.__init__()
self.dropout = nn.Dropout(dropout)
self.norm = LayerNormalization()
def forward(self,x,sublayer):
return x + self.dropout(sublayer(self.norm(x)))
class EncoderBlock(nn.Module):
def __init__(self,self_attention_block:MultiHeadAttentionBlock,feed_forward_block:FeedForwardBlock,dropout:float)->None:
super().__init__()
self.self_attention_block = self_attention_block
self.feed_forward_block = feed_forward_block
self.residual_connection = nn.ModuleList([ResidualConnection(dropout)for _ in range(2)])
#
def forward(self,x,src_mask):
x = self.residual_connections[0](x,lambda x: self.self_attention_block(x,x,x,src_mask))
x = self.residual_connections[1](x,self.feed_forward_block)
return x
class Encoder(nn.Module):
def __init__(self,layers:nn.Module)->None:
super().__init__()
self.layers = layers
self.norm = LayerNormalization()
def forward(self,x,mask):
for layer in self.layers:
x = layer(x,mask)
return self.norm(x)
class Decoderblock(nn.Module):
def __init__(self,self_attention_block:MultiHeadAttentionBlock,cross_attention_block:MultiHeadAttentionBlock,feed_forward_block:FeedForwardBlock,dropout:float)->None:
super().__init__()
self.self_attention_block = self_attention_block
self.cross_attention_block = cross_attention_block
self.feed_forward_block = feed_forward_block
self.residual_connections = nn.Module([ResidualConnection(dropout) for _ in range(3)])
#input of the decoder, output of the encoder, mask applied to the encoder, mask applied to the decoder
def forward(self,x,encoder_output,src_mask,target_mask):
x = self.residual_connections[0](x,lambda x: self.self_attention_block(x,x,x,target_mask))
x = self.residual_connections[1](x,lambda x:self.cross_attention_block(x,encoder_output,encoder_output,src_mask))
x = self.residual_connections[2](x,self.feed_forward_block)
return x
class Decoder(nn.Module):
def __init__(self,layers:nn.Module)->None:
super().__init__()
self.layers = layers
self.norm = LayerNormalization()
def forward(self,x,encoder_output,src_mask,target_mask):
for layer in self.layers:
x = layer(x,encoder_output,src_mask,target_mask)
return self.norm(x)
#convert the embedding to a position in the vocabulary
class ProjectionLayer(nn.Module):
def __init__(self,d_model:int,vocab_size:int)->None:
super().__init__()
self.proj = nn.Linear(d_model,vocab_size)
def forward(self,x):
#(batch,seq_len,d_model) ---> (batch,seq_len,vocab_size)
return torch.log_softmax(self.proj(x),dim=-1)