Pytorch のsummary表示はどうしてダブっていますか？

python
1
2# モデルクラスの定義
3class cifar10_cnn(nn.Module):
4  def __init__(self, num_classes):
5    super(cifar10_cnn,self).__init__()
6    self.conv1 = nn.Conv2d(3, 32, 3, padding=(1,1), padding_mode='replicate')
7    self.conv2 = nn.Conv2d(32, 32, 3)
8    self.conv3 = nn.Conv2d(32, 64, 3, padding=(1,1), padding_mode='replicate')
9    self.conv4 = nn.Conv2d(64, 64, 3)
10    self.relu = nn.ReLU(inplace=True)
11    self.dropout1 = nn.Dropout(0.25)
12    self.dropout2 = nn.Dropout(0.5)
13    self.maxpool = nn.MaxPool2d((2,2))
14    self.classifier1 = nn.Linear(2304, 512)
15    self.classifier2 = nn.Linear(512, num_classes)
16
17    self.features = nn.Sequential(
18        self.conv1,
19        self.relu,
20        self.conv2,
21        self.relu,
22        self.maxpool,
23        self.dropout1,
24        self.conv3,
25        self.relu,
26        self.conv4,
27        self.relu,
28        self.dropout1,
29        self.maxpool)
30    
31    self.classifier = nn.Sequential(
32       self.classifier1,
33       self.relu,
34       self.dropout2,
35       self.classifier2)
36
37  def forward(self, x):
38    x = self.features(x)
39    x = x.view(x.size(0), -1)
40    x = self.classifier(x)
41    return x
42  
43# GPUの確認
44device = 'cuda' if torch.cuda.is_available() else 'cpu'
45print("\n device : ", device)
46
47# モデルインスタンスの生成とGPUの割り当て
48net = cifar10_cnn(num_classes).to(device)
49
50criterion = nn.CrossEntropyLoss()
51optimizer = optim.Adam(net.parameters())
52
53# モデルのサマリー表示
54from torchsummary import summary
55
56print("\n Model summary : summary(net,(3,32,32))\n")#(3,32,32) | 3: 入力channel数; (32,32):入力面積
57summary(net,(3,32,32))
58
59**以下はsummary(net,(3,32,32))からの表示**　
60👹👹Conv2dは2つずつ表示され、総数は定義した分の２倍になってしまいます
61原因は何でしょうか？
62
63---
64Model summary : summary(net,(3,32,32))
65
66----------------------------------------------------------------
67        Layer (type)               Output Shape         Param #
68================================================================
69            Conv2d-1           [-1, 32, 32, 32]             896
70            Conv2d-2           [-1, 32, 32, 32]             896
71              ReLU-3           [-1, 32, 32, 32]               0
72              ReLU-4           [-1, 32, 32, 32]               0
73              ReLU-5           [-1, 32, 32, 32]               0
74            Conv2d-6           [-1, 32, 30, 30]           9,248
75            Conv2d-7           [-1, 32, 30, 30]           9,248
76              ReLU-8           [-1, 32, 30, 30]               0
77              ReLU-9           [-1, 32, 30, 30]               0
78             ReLU-10           [-1, 32, 30, 30]               0
79        MaxPool2d-11           [-1, 32, 15, 15]               0
80        MaxPool2d-12           [-1, 32, 15, 15]               0
81          Dropout-13           [-1, 32, 15, 15]               0
82          Dropout-14           [-1, 32, 15, 15]               0
83           Conv2d-15           [-1, 64, 15, 15]          18,496
84           Conv2d-16           [-1, 64, 15, 15]          18,496
85             ReLU-17           [-1, 64, 15, 15]               0
86             ReLU-18           [-1, 64, 15, 15]               0
87             ReLU-19           [-1, 64, 15, 15]               0
88           Conv2d-20           [-1, 64, 13, 13]          36,928
89           Conv2d-21           [-1, 64, 13, 13]          36,928
90             ReLU-22           [-1, 64, 13, 13]               0
91             ReLU-23           [-1, 64, 13, 13]               0
92             ReLU-24           [-1, 64, 13, 13]               0
93          Dropout-25           [-1, 64, 13, 13]               0
94          Dropout-26           [-1, 64, 13, 13]               0
95        MaxPool2d-27             [-1, 64, 6, 6]               0
96        MaxPool2d-28             [-1, 64, 6, 6]               0
97           Linear-29                  [-1, 512]       1,180,160
98           Linear-30                  [-1, 512]       1,180,160
99             ReLU-31                  [-1, 512]               0
100             ReLU-32                  [-1, 512]               0
101             ReLU-33                  [-1, 512]               0
102          Dropout-34                  [-1, 512]               0
103          Dropout-35                  [-1, 512]               0
104           Linear-36                   [-1, 10]           5,130
105           Linear-37                   [-1, 10]           5,130
106================================================================
107Total params: 2,501,716
108Trainable params: 2,501,716
109Non-trainable params: 0
110----------------------------------------------------------------
111Input size (MB): 0.01
112Forward/backward pass size (MB): 3.76
113Params size (MB): 9.54
114Estimated Total Size (MB): 13.31