torch數據維度

 

3,1 表示最外層括號中有3條數據，2個逗號

 

torch的數據維度表示，從外向裏表示，[3,1]，表示外面有3行1列

 

import torch
    import torch.nn as nn
    import torch.nn.functional as F

    data = torch.randn(3,1,6)
    data2 = torch.randn(3, 1)
    print(data2.size())
    print(data2)

    data2 = torch.randn(1, 3)
    print(data2.size())
    print(data2)

 

torch.Size([3, 1])
tensor([[ 1.5015],
        [-1.8209],
        [ 0.9194]])
torch.Size([1, 3])
tensor([[ 1.4510, -1.3262, -0.5420]])

 

target的第2維和data的第3維數量要保持一致：不然報錯：

import torch
    import torch.nn as nn
    import torch.nn.functional as F

    data = torch.randn(3,3,2)

    target = torch.tensor([[1,1],[0,1],[0,1]])
    print(data.size(),target.size())

    n = data.size(0)
    c = data.size(1)
    out_size = (n,) + data.size()[2:]
    if target.size()[1:] != data.size()[2:]:
        raise ValueError('Expected target size {}, got {}'.format(
            out_size, target.size()))
    #
    # print('data:', data.size(),target.size())
    entropy_out = F.cross_entropy(data, target)
    print('entropy_out:', entropy_out, target.size())