Update app.py
Browse files
app.py
CHANGED
@@ -5,6 +5,29 @@ import matplotlib.pyplot as plt
|
|
5 |
import numpy as np
|
6 |
from torch import nn
|
7 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8 |
path = hf_hub_download('huggan/ArtGAN', 'ArtGAN.pt')
|
9 |
model = torch.load(path, map_location=torch.device('cpu'))
|
10 |
device = 'cuda' if torch.cuda.is_available() else 'cpu'
|
|
|
5 |
import numpy as np
|
6 |
from torch import nn
|
7 |
|
8 |
+
class Generator(nn.Module):
|
9 |
+
def __init__(self):
|
10 |
+
super(Generator, self).__init__()
|
11 |
+
self.main = nn.Sequential(
|
12 |
+
nn.ConvTranspose2d(100, 64 * 8, 4, 1, 0, bias=False),
|
13 |
+
nn.BatchNorm2d(64 * 8),
|
14 |
+
nn.ReLU(True),
|
15 |
+
nn.ConvTranspose2d(64 * 8, 64 * 4, 4, 2, 1, bias=False),
|
16 |
+
nn.BatchNorm2d(64 * 4),
|
17 |
+
nn.ReLU(True),
|
18 |
+
nn.ConvTranspose2d(64 * 4, 64 * 2, 4, 2, 1, bias=False),
|
19 |
+
nn.BatchNorm2d(64 * 2),
|
20 |
+
nn.ReLU(True),
|
21 |
+
nn.ConvTranspose2d(64 * 2, 64, 4, 2, 1, bias=False),
|
22 |
+
nn.BatchNorm2d(64),
|
23 |
+
nn.ReLU(True),
|
24 |
+
nn.ConvTranspose2d(64, 3, 4, 2, 1, bias=False),
|
25 |
+
nn.Tanh()
|
26 |
+
)
|
27 |
+
|
28 |
+
def forward(self, input):
|
29 |
+
return self.main(input)
|
30 |
+
|
31 |
path = hf_hub_download('huggan/ArtGAN', 'ArtGAN.pt')
|
32 |
model = torch.load(path, map_location=torch.device('cpu'))
|
33 |
device = 'cuda' if torch.cuda.is_available() else 'cpu'
|