【PyTorch学习4】张量拼接

1. torch.cat(data,dim)

data11 = torch.randint(0,10,[2,3,2])  # torch.Size([2, 3, 2])
data12 = torch.randint(10,20,[2,3,2])  # torch.Size([2, 3, 2])
data13 = torch.cat([data11,data12],0)  # 按照第0维拼接 torch.Size([4, 3, 2])
data14 = torch.cat([data11,data12],1)  # 按照第1维拼接 torch.Size([2, 6, 2])
data15 = torch.cat([data11,data12],2)  # 按照第2维拼接 torch.Size([2, 3, 4])

2. torch.stack(data,dim)

torch.manual_seed(1)
data21 = torch.randint(0,10,[3,3])  # torch.Size([3, 3])
'''
tensor([[5, 9, 4],
        [8, 3, 3],
        [1, 1, 9]])
'''
data22 = torch.randint(10,20,[3,3])  # torch.Size([3, 3])
'''
tensor([[12, 18, 19],
        [16, 13, 13],
        [10, 12, 11]])
'''
data23 = torch.stack([data21,data22],0)  # 按照第0维拼接 torch.Size([2, 3, 3])
'''
tensor([[[ 5,  9,  4],
         [ 8,  3,  3],
         [ 1,  1,  9]],

        [[12, 18, 19],
         [16, 13, 13],
         [10, 12, 11]]])
'''
data24 = torch.stack([data21,data22],1)  # 按照第1维拼接 torch.Size([3, 2, 3])
'''
tensor([[[ 5,  9,  4],
         [12, 18, 19]],

        [[ 8,  3,  3],
         [16, 13, 13]],

        [[ 1,  1,  9],
         [10, 12, 11]]])
'''
data25 = torch.stack([data21,data22],2)  # 按照第2维拼接 torch.Size([3, 3, 2])
'''
tensor([[[ 5, 12],
         [ 9, 18],
         [ 4, 19]],

        [[ 8, 16],
         [ 3, 13],
         [ 3, 13]],

        [[ 1, 10],
         [ 1, 12],
         [ 9, 11]]])
'''