-
Notifications
You must be signed in to change notification settings - Fork 17
/
Copy pathmodel.py
191 lines (150 loc) · 7.46 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
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
188
189
190
191
import torch
import torch.nn as nn
import torch.nn.functional as F
from .conv_layers import MaskedConv2d, CroppedConv2d
class CausalBlock(nn.Module):
def __init__(self, in_channels, out_channels, kernel_size, data_channels):
super(CausalBlock, self).__init__()
self.split_size = out_channels
self.v_conv = CroppedConv2d(in_channels,
2 * out_channels,
(kernel_size // 2 + 1, kernel_size),
padding=(kernel_size // 2 + 1, kernel_size // 2))
self.v_fc = nn.Conv2d(in_channels,
2 * out_channels,
(1, 1))
self.v_to_h = nn.Conv2d(2 * out_channels,
2 * out_channels,
(1, 1))
self.h_conv = MaskedConv2d(in_channels,
2 * out_channels,
(1, kernel_size),
mask_type='A',
data_channels=data_channels,
padding=(0, kernel_size // 2))
self.h_fc = MaskedConv2d(out_channels,
out_channels,
(1, 1),
mask_type='A',
data_channels=data_channels)
def forward(self, image):
v_out, v_shifted = self.v_conv(image)
v_out += self.v_fc(image)
v_out_tanh, v_out_sigmoid = torch.split(v_out, self.split_size, dim=1)
v_out = torch.tanh(v_out_tanh) * torch.sigmoid(v_out_sigmoid)
h_out = self.h_conv(image)
v_shifted = self.v_to_h(v_shifted)
h_out += v_shifted
h_out_tanh, h_out_sigmoid = torch.split(h_out, self.split_size, dim=1)
h_out = torch.tanh(h_out_tanh) * torch.sigmoid(h_out_sigmoid)
h_out = self.h_fc(h_out)
return v_out, h_out
class GatedBlock(nn.Module):
def __init__(self, in_channels, out_channels, kernel_size, data_channels):
super(GatedBlock, self).__init__()
self.split_size = out_channels
self.v_conv = CroppedConv2d(in_channels,
2 * out_channels,
(kernel_size // 2 + 1, kernel_size),
padding=(kernel_size // 2 + 1, kernel_size // 2))
self.v_fc = nn.Conv2d(in_channels,
2 * out_channels,
(1, 1))
self.v_to_h = MaskedConv2d(2 * out_channels,
2 * out_channels,
(1, 1),
mask_type='B',
data_channels=data_channels)
self.h_conv = MaskedConv2d(in_channels,
2 * out_channels,
(1, kernel_size),
mask_type='B',
data_channels=data_channels,
padding=(0, kernel_size // 2))
self.h_fc = MaskedConv2d(out_channels,
out_channels,
(1, 1),
mask_type='B',
data_channels=data_channels)
self.h_skip = MaskedConv2d(out_channels,
out_channels,
(1, 1),
mask_type='B',
data_channels=data_channels)
self.label_embedding = nn.Embedding(10, 2*out_channels)
def forward(self, x):
v_in, h_in, skip, label = x[0], x[1], x[2], x[3]
label_embedded = self.label_embedding(label).unsqueeze(2).unsqueeze(3)
v_out, v_shifted = self.v_conv(v_in)
v_out += self.v_fc(v_in)
v_out += label_embedded
v_out_tanh, v_out_sigmoid = torch.split(v_out, self.split_size, dim=1)
v_out = torch.tanh(v_out_tanh) * torch.sigmoid(v_out_sigmoid)
h_out = self.h_conv(h_in)
v_shifted = self.v_to_h(v_shifted)
h_out += v_shifted
h_out += label_embedded
h_out_tanh, h_out_sigmoid = torch.split(h_out, self.split_size, dim=1)
h_out = torch.tanh(h_out_tanh) * torch.sigmoid(h_out_sigmoid)
# skip connection
skip = skip + self.h_skip(h_out)
h_out = self.h_fc(h_out)
# residual connections
h_out = h_out + h_in
v_out = v_out + v_in
return {0: v_out, 1: h_out, 2: skip, 3: label}
class PixelCNN(nn.Module):
def __init__(self, cfg):
super(PixelCNN, self).__init__()
DATA_CHANNELS = 3
self.hidden_fmaps = cfg.hidden_fmaps
self.color_levels = cfg.color_levels
self.causal_conv = CausalBlock(DATA_CHANNELS,
cfg.hidden_fmaps,
cfg.causal_ksize,
data_channels=DATA_CHANNELS)
self.hidden_conv = nn.Sequential(
*[GatedBlock(cfg.hidden_fmaps, cfg.hidden_fmaps, cfg.hidden_ksize, DATA_CHANNELS) for _ in range(cfg.hidden_layers)]
)
self.label_embedding = nn.Embedding(10, self.hidden_fmaps)
self.out_hidden_conv = MaskedConv2d(cfg.hidden_fmaps,
cfg.out_hidden_fmaps,
(1, 1),
mask_type='B',
data_channels=DATA_CHANNELS)
self.out_conv = MaskedConv2d(cfg.out_hidden_fmaps,
DATA_CHANNELS * cfg.color_levels,
(1, 1),
mask_type='B',
data_channels=DATA_CHANNELS)
def forward(self, image, label):
count, data_channels, height, width = image.size()
v, h = self.causal_conv(image)
_, _, out, _ = self.hidden_conv({0: v,
1: h,
2: image.new_zeros((count, self.hidden_fmaps, height, width), requires_grad=True),
3: label}).values()
label_embedded = self.label_embedding(label).unsqueeze(2).unsqueeze(3)
# add label bias
out += label_embedded
out = F.relu(out)
out = F.relu(self.out_hidden_conv(out))
out = self.out_conv(out)
out = out.view(count, self.color_levels, data_channels, height, width)
return out
def sample(self, shape, count, label=None, device='cuda'):
channels, height, width = shape
samples = torch.zeros(count, *shape).to(device)
if label is None:
labels = torch.randint(high=10, size=(count,)).to(device)
else:
labels = (label*torch.ones(count)).to(device).long()
with torch.no_grad():
for i in range(height):
for j in range(width):
for c in range(channels):
unnormalized_probs = self.forward(samples, labels)
pixel_probs = torch.softmax(unnormalized_probs[:, :, c, i, j], dim=1)
sampled_levels = torch.multinomial(pixel_probs, 1).squeeze().float() / (self.color_levels - 1)
samples[:, c, i, j] = sampled_levels
return samples