Remove in_conv in ResNet.

hsaghir · Dec 9, 2018 · 3374f19 · 3374f19
1 parent 73922d0
commit 3374f19
Showing 1 changed file with 1 addition and 5 deletions.
diff --git a/models/resnet/resnet.py b/models/resnet/resnet.py
@@ -1,4 +1,3 @@
-import torch
 import torch.nn as nn
 import torch.nn.functional as F
 
@@ -18,8 +17,7 @@ class ResNet(nn.Module):
     """
     def __init__(self, in_channels, mid_channels, out_channels, num_blocks, kernel_size, padding):
         super(ResNet, self).__init__()
-        self.in_norm = nn.BatchNorm2d(in_channels, affine=False)
-        self.in_conv = WNConv2d(2 * in_channels, mid_channels, kernel_size, padding, bias=True)
+        self.in_conv = WNConv2d(in_channels, mid_channels, kernel_size, padding, bias=True)
         self.in_skip = WNConv2d(mid_channels, mid_channels, kernel_size=1, padding=0, bias=True)
 
         self.blocks = nn.ModuleList([ResidualBlock(mid_channels, mid_channels)
@@ -31,8 +29,6 @@ def __init__(self, in_channels, mid_channels, out_channels, num_blocks, kernel_s
         self.out_conv = WNConv2d(mid_channels, out_channels, kernel_size=1, padding=0, bias=True)
 
     def forward(self, x):
-        x = self.in_norm(x)
-        x = F.relu(torch.cat((x, -x), dim=1))
         x = self.in_conv(x)
         x_skip = self.in_skip(x)