pytorch forward两个参数实例

Python  /  管理员 发布于 6年前   436

以channel Attention Block为例子

class CAB(nn.Module):   def __init__(self, in_channels, out_channels):    super(CAB, self).__init__()    self.global_pooling = nn.AdaptiveAvgPool2d(output_size=1)    self.conv1 = nn.Conv2d(in_channels, out_channels, kernel_size=1, stride=1, padding=0)    self.relu = nn.ReLU()    self.conv2 = nn.Conv2d(out_channels, out_channels, kernel_size=1, stride=1, padding=0)    self.sigmod = nn.Sigmoid()   def forward(self, x):    x1, x2 = x # high, low    x = torch.cat([x1,x2],dim=1)    x = self.global_pooling(x)    x = self.conv1(x)    x = self.relu(x)    x = self.conv2(x)    x = self.sigmod(x)    x2 = x * x2    res = x2 + x1    return res

以上这篇pytorch forward两个参数实例就是小编分享给大家的全部内容了，希望能给大家一个参考，也希望大家多多支持。