Tensor and autograd

1. Different t.Tensor and the t.tensor

　　　t.Tensor (size) can create a form for the size of the direct tensor

　　　t.tensor () requires t.tensor ([1, 2]) to create what is regardless of the input type, t.tensor will copy data, do not share memory

2. resize () and view () different

　　a resize () can be modified tensor size, and view () can adjust the shape

3.　　a[0:1, :2].size()=torch.Size([1, 2])

　　a[0, :2].size()=torch.Size([2])

4. Parametersdim

Suppose that the shape of the input (m, n, k)

• • If you specify dim = 0, is the shape of the output (1, n, k) or (n, k)
  • If the specified dim = 1, is the shape of the output (m, 1, k) or (m, k)
  • If you specify dim = 2, it is the shape of the output (m, n, 1) or (m, n)

Is there a "1", depending on the size of the parameters keepdim, keepdim=Truewill remain dimension 1.

Not all functions are in line with changes in the way this shapes, such ascumsum

5. torch.mm () matrix multiplication

　　torch.mul () bit corresponding to a matrix multiplication