File size: 3,744 Bytes
b993f12
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import torch 
from torch import nn
from attention import SelfAttention
from torch.nn import functional as F


class VAE_ResidualBlock(nn.Module):
    def __init__(self, in_channels, out_channels):
        super().__init__()
        self.grpnorm_1=nn.GroupNorm(32, in_channels)
        self.conv_1=nn.Conv2d(in_channels, out_channels, kernel_size=3, padding=1)

        self.grpnorm_2=nn.GroupNorm(32, out_channels)
        self.conv_2=nn.Conv2d(out_channels, out_channels, kernel_size=3, padding=1)

        if in_channels == out_channels:
            self.residual_layer=nn.Identity()
        else:
            self.residual_layer=nn.Conv2d(in_channels, out_channels, kernel_size=1, padding=0)
    def forward(self, x):
        residue=x

        x=self.grpnorm_1(x)
        x=F.silu(x)

        x=self.conv_1(x)

        x=self.grpnorm_2(x)
        x=F.silu(x)

        x=self.conv_2(x)

        return x+self.residual_layer(residue)

class VAE_AttentionBlock(nn.Module):
    def __init__(self, channels):
        super().__init__()
        self.grpnorm=nn.GroupNorm(32, channels)
        self.attention=SelfAttention(1, channels)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        # x: (Batch_Size, Features, Height, Width)
        residue=x

        # (Batch_Size, Features, Height, Width) -> (Batch_Size, Features, Height, Width)
        x=self.grpnorm(x)
        n, c, h, w=x.shape

        # (Batch_Size, Features, Height, Width) -> (Batch_Size, Features, Height * Width)
        x=x.view((n,c,h*w))

        # (Batch_Size, Features, Height * Width) -> (Batch_Size, Height * Width, Features)
        x=x.transpose(-1, -2)

        # (Batch_Size, Height * Width, Features) -> (Batch_Size, Height * Width, Features) -> (Batch_Size, Height * Width, Features) 
        x=self.attention(x)

        # (Batch_Size, Height * Width, Features)  -> (Batch_Size, Features, Height * Width)
        x=x.transpose(-1, -2)

        # (Batch_Size, Features, Height , Width)
        x=x.view((n, c, h, w))

        x+=residue

        return x

class VAE_Decoder(nn.Sequential):
    def __init__(self):
        super().__init__(
            nn.Conv2d(4, 4, kernel_size=1, padding=0),

            nn.Conv2d(4, 512, kernel_size=3, padding=1),

            VAE_ResidualBlock(512, 512),

            VAE_AttentionBlock(512),

            VAE_ResidualBlock(512, 512),
            VAE_ResidualBlock(512, 512),
            VAE_ResidualBlock(512, 512),
            VAE_ResidualBlock(512, 512),

            nn.Upsample(scale_factor=2),

            nn.Conv2d(512, 512, kernel_size=3, padding=1),

            VAE_ResidualBlock(512, 512),
            VAE_ResidualBlock(512, 512),
            VAE_ResidualBlock(512, 512),

            nn.Upsample(scale_factor=2),
            
            nn.Conv2d(512, 512, kernel_size=3, padding=1),

            VAE_ResidualBlock(512, 256),
            VAE_ResidualBlock(256, 256),
            VAE_ResidualBlock(256, 256),

            nn.Upsample(scale_factor=2),

            nn.Conv2d(256, 256, kernel_size=3, padding=1),

            VAE_ResidualBlock(256, 128),
            VAE_ResidualBlock(128, 128),
            VAE_ResidualBlock(128, 128),

            nn.GroupNorm(32, 128),

            nn.SiLU(),

            nn.Conv2d(128, 3, kernel_size=3, padding=1),
        )

    def forward(self, x):
        x/=0.18215

        for module in self:
            x=module(x)

        return x

if __name__ == "__main__":
    model = VAE_Decoder()
    model.eval()

    # Create a dummy input tensor: (batch_size=1, channels=4, height=16, width=16)
    x = torch.randn(1, 4, 8, 8)

    with torch.no_grad():
        output = model(x)

    print("Input shape :", x.shape)
    print("Output shape:", output.shape)