-
Notifications
You must be signed in to change notification settings - Fork 1
/
model.py
159 lines (144 loc) · 7.91 KB
/
model.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
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.nn.utils.rnn as rnn_util
from torch.nn.utils.rnn import pad_packed_sequence, pack_padded_sequence
from transformer import *
import conformer as cf
from convolution_module import Conv1dSubampling
from pooling_layers import *
#
class Transformer_E2E_LID(nn.Module):
def __init__(self, input_dim, feat_dim,
d_k, d_v, d_ff, n_heads=8,
dropout=0.1,n_lang=3, max_seq_len=300,
device=torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')):
super(Transformer_E2E_LID, self).__init__()
# self.subsample = Conv1dSubampling()
self.transform = nn.Linear(input_dim, feat_dim)
self.layernorm1 = LayerNorm(feat_dim)
self.pos_encoding = PositionalEncoding(max_seq_len=max_seq_len, features_dim=feat_dim, device=device)
self.layernorm2 = LayerNorm(feat_dim)
self.d_model = feat_dim*n_heads
self.n_heads = n_heads
self.attention_block1 = EncoderBlock(self.d_model, d_k, d_v, d_ff, n_heads, dropout=dropout)
self.attention_block2 = EncoderBlock(self.d_model, d_k, d_v, d_ff, n_heads, dropout=dropout)
self.attention_block3 = EncoderBlock(self.d_model, d_k, d_v, d_ff, n_heads, dropout=dropout)
self.attention_block4 = EncoderBlock(self.d_model, d_k, d_v, d_ff, n_heads, dropout=dropout)
self.fc1 = nn.Linear(self.d_model * 2, self.d_model)
self.fc2 = nn.Linear(self.d_model, self.d_model)
self.fc3 = nn.Linear(self.d_model, n_lang)
self.sigmoid = nn.Sigmoid()
def forward(self, x, seq_len, atten_mask):
batch_size = x.size(0)
output = self.transform(x) #x [B, T, input_dim] => [B, T feat_dim]
output = self.layernorm1(output)
output = self.pos_encoding(output, seq_len)
output = self.layernorm2(output)
output = output.unsqueeze(1).repeat(1, self.n_heads, 1, 1)
output = output.transpose(1, 2).contiguous().view(batch_size, -1, self.d_model)
output, _ = self.attention_block1(output, atten_mask)
output, _ = self.attention_block2(output, atten_mask)
output, _ = self.attention_block3(output, atten_mask)
output, _ = self.attention_block4(output, atten_mask)
stats = torch.cat((output.mean(dim=1), output.std(dim=1)), dim=1)
# print(stats.size())
output = F.relu(self.fc1(stats))
output = F.relu(self.fc2(output))
output = self.fc3(output)
return output
class X_Transformer_E2E_LID(nn.Module):
def __init__(self, input_dim, feat_dim,
d_k, d_v, d_ff, n_heads=4,
dropout=0.1, n_lang=3, max_seq_len=140,
device=torch.device('cuda:0' if torch.cuda.is_available() else 'cpu')):
super(X_Transformer_E2E_LID, self).__init__()
self.input_dim = input_dim
self.feat_dim = feat_dim
self.device = device
self.dropout = nn.Dropout(p=dropout)
self.tdnn1 = nn.Conv1d(in_channels=input_dim, out_channels=512, kernel_size=5, dilation=1)
self.bn1 = nn.BatchNorm1d(512, momentum=0.1, affine=False)
self.tdnn2 = nn.Conv1d(in_channels=512, out_channels=512, kernel_size=5, dilation=2)
self.bn2 = nn.BatchNorm1d(512, momentum=0.1, affine=False)
self.tdnn3 = nn.Conv1d(in_channels=512, out_channels=512, kernel_size=1, dilation=1)
self.bn3 = nn.BatchNorm1d(512, momentum=0.1, affine=False)
self.fc_xv = nn.Linear(1024, feat_dim)
self.layernorm1 = LayerNorm(feat_dim)
self.pos_encoding = PositionalEncoding(max_seq_len=max_seq_len, features_dim=feat_dim, device=device)
self.layernorm2 = LayerNorm(feat_dim)
self.d_model = feat_dim * n_heads
self.n_heads = n_heads
self.attention_block1 = EncoderBlock(self.d_model, d_k, d_v, d_ff, n_heads, dropout=dropout)
self.attention_block2 = EncoderBlock(self.d_model, d_k, d_v, d_ff, n_heads, dropout=dropout)
# self.attention_block3 = EncoderBlock(self.d_model, d_k, d_v, d_ff, n_heads, dropout=dropout)
# self.attention_block4 = EncoderBlock(self.d_model, d_k, d_v, d_ff, n_heads, dropout=dropout)
# self.attention_block5 = EncoderBlock(self.d_model, d_k, d_v, d_ff, n_heads, dropout=dropout)
# self.attention_block6 = EncoderBlock(self.d_model, d_k, d_v, d_ff, n_heads, dropout=dropout)
self.fc1 = nn.Linear(self.d_model * 2, self.d_model)
self.fc2 = nn.Linear(self.d_model, self.d_model)
self.fc3 = nn.Linear(self.d_model, n_lang)
def forward(self, x, seq_len, atten_mask, eps=1e-5):
batch_size = x.size(0)
T_len = x.size(1)
x = self.dropout(x)
x = x.view(batch_size * T_len, -1, self.input_dim).transpose(-1, -2)
x = self.bn1(F.relu(self.tdnn1(x)))
x = self.bn2(F.relu(self.tdnn2(x)))
x = self.bn3(F.relu(self.tdnn3(x)))
if self.training:
shape = x.size()
noise = torch.Tensor(shape)
noise = noise.type_as(x)
torch.randn(shape, out=noise)
x += noise * eps
stats = torch.cat((x.mean(dim=2), x.std(dim=2)), dim=1)
# print("pooling", stats.size())
embedding = self.fc_xv(stats)
embedding = embedding.view(batch_size, T_len, self.feat_dim)
output = self.layernorm1(embedding)
output = self.pos_encoding(output, seq_len)
output = self.layernorm2(output)
output = output.unsqueeze(1).repeat(1, self.n_heads, 1, 1)
output = output.transpose(1, 2).contiguous().view(batch_size, -1, self.d_model)
output, _ = self.attention_block1(output, atten_mask)
output, _ = self.attention_block2(output, atten_mask)
# output, _ = self.attention_block3(output, atten_mask)
# output, _ = self.attention_block4(output, atten_mask)
stats = torch.cat((output.mean(dim=1), output.std(dim=1)), dim=1)
output = F.relu(self.fc1(stats))
output = F.relu(self.fc2(output))
output = self.fc3(output)
return output
class Conformer(nn.Module):
def __init__(self,input_dim, feat_dim, d_k, d_v, n_heads, d_ff, max_len, dropout, device, n_lang):
super(Conformer, self).__init__()
self.conv_subsample = Conv1dSubampling(in_channels=input_dim, out_channels= input_dim)
self.transform = nn.Linear(input_dim, feat_dim)
self.dropout = nn.Dropout(dropout)
self.d_model = feat_dim * n_heads
self.layernorm1 = LayerNorm(feat_dim)
self.n_heads = n_heads
self.attention_block1 = cf.ConformerEncoder(self.d_model, d_k, d_v, d_ff, n_heads, dropout, max_len, device)
self.attention_block2 = cf.ConformerEncoder(self.d_model, d_k, d_v, d_ff, n_heads, dropout, max_len, device)
self.attention_block3 = cf.ConformerEncoder(self.d_model, d_k, d_v, d_ff, n_heads, dropout, max_len, device)
self.attention_block4 = cf.ConformerEncoder(self.d_model, d_k, d_v, d_ff, n_heads, dropout, max_len, device)
self.fc1 = nn.Linear(self.d_model * 2, self.d_model)
self.fc2 = nn.Linear(self.d_model, self.d_model)
self.fc3 = nn.Linear(self.d_model, n_lang)
def forward(self, x, atten_mask):
batch_size = x.size(0)
output = self.transform(x) # x [B, T, input_dim] => [B, T feat_dim]
output = self.layernorm1(output)
output = output.unsqueeze(1).repeat(1, self.n_heads, 1, 1)
output = output.transpose(1, 2).contiguous().view(batch_size, -1, self.d_model)
output, _ = self.attention_block1(output, atten_mask)
output, _ = self.attention_block2(output, atten_mask)
output, _ = self.attention_block3(output, atten_mask)
output, _ = self.attention_block4(output, atten_mask)
stats = torch.cat((output.mean(dim=1), output.std(dim=1)), dim=1)
# print(stats.size())
output = F.relu(self.fc1(stats))
output = F.relu(self.fc2(output))
output = self.fc3(output)
return output