Dear:
i understand tensor shape torch.Size([1])
and scalar shape torch.Size([])
, but
what does torch.Size([0])
means generated by torch.randn(0)
, is it a bug??
In [4]: a=torch.randn(0)
In [5]: a
Out[5]: tensor([])
In [8]: a.shape
Out[8]: torch.Size([0])
In [9]: a=torch.tensor(0)
In [10]: a.shape
Out[10]: torch.Size([])
In [11]: a=torch.tensor([0.0])
In [12]: a.shape
Out[12]: torch.Size([1])