一. tensor.T解析
1. 说明:
官网:tensor.T,tensor.T表示对tensor张量进行转置,解析如下图所示:
2. 代码例子
2.1 一维张量转置(与原张量相同)
a = torch.randn(3)
a,a.T
结果如下所示:
(tensor([ 1.62, -0.06, -1.31]), tensor([ 1.62, -0.06, -1.31]))
2.2 二维张量转置
a = torch.randn(2,3)
a,a.T #a.T结果由两行三列变成三行两列
结果如下所示:
(tensor([[-1.14, 0.35, -1.82],
[ 2.33, 1.34, 0.11]]),
tensor([[-1.14, 2.33],
[ 0.35, 1.34],
[-1.82, 0.11]]))
2.3 三维张量转置
a = torch.randn(2,3,3)
a,a.T
结果如下所示:
(tensor([[[-0.57, -0.99, -1.53],
[ 0.03, 0.56, 1.36],
[ 0.89, -0.46, 1.48]],
[[-0.31, -0.62, -0.04],
[ 1.39, -2.48, -1.09],
[-0.28, -1.24, 0.47]]]),
tensor([[[-0.57, -0.31],
[ 0.03, 1.39],
[ 0.89, -0.28]],
[[-0.99, -0.62],
[ 0.56, -2.48],
[-0.46, -1.24]],
[[-1.53, -0.04],
[ 1.36, -1.09],
[ 1.48, 0.47]]]))