pyrotch下张量的拼接和线性变换

导包

import torch
import torch.nn as nn

创建神经网络模型类：

class _netG(nn.Module):
    def  __init__(self):
        super(_netG,self).__init__()
        self.fc1 = nn.Linear(2048,1024)线性变换

    def forward(self,x):

        x_1 = self.fc1(x)
        print(x_1.shape)

测试代码：

if __name__ == '__main__':

    gt = torch.rand(4, 1024).cuda()
    pre_end = torch.rand(4, 1024).cuda()
    print(gt.shape, pre_end.shape)

    a = torch.cat((gt, pre_end), dim=1).cuda()#向量拼接
    print(a.shape)
    mynet = _netG().cuda()
    mynet(a)