-
Notifications
You must be signed in to change notification settings - Fork 1
/
module.py
35 lines (30 loc) · 825 Bytes
/
module.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
import torch
import torch.nn as nn
class SharedMLP(nn.Module):
def __init__(self, in_dim, out_dim):
super().__init__()
self.main = nn.Sequential(
nn.Conv1d(in_dim, out_dim, 1),
nn.BatchNorm1d(out_dim),
nn.LeakyReLU()
)
def forward(self, x):
out = self.main(x)
return out
class LinearMLP(nn.Module):
def __init__(self, in_dim, out_dim):
super().__init__()
self.main = nn.Sequential(
nn.Linear(in_dim, out_dim),
nn.BatchNorm1d(out_dim),
nn.LeakyReLU()
)
def forward(self, x):
out = self.main(x)
return out
if __name__ == "__main__":
x = torch.randn(100, 2, 199)
Net = SharedMLP(2, 10)
out = Net(x)
print(x.shape)
print(out.shape)