How to alternatively concatenate pytorch tensors?

Question

Pytorch provides API to concatenate tensors, like cat, stack. But does it provide any API to concatenate pytorch tensors alternatively?

For example， enter image description here

suppose input1.shape = C*H*W, a1.shape = H\*W, and output.shape = (3C)*H*W

This can be achieved using a loop, but I am wondering if any Pytorch API can do this

Where is a1 in your illustrative example?

Hamzah Al-Qadasi
– Hamzah Al-Qadasi

2022-03-26 14:02:09 +00:00
Commented Mar 26, 2022 at 14:02 — Hamzah Al-Qadasi
– Hamzah Al-Qadasi, Commented Mar 26, 2022 at 14:02
a1 = the first line of input1. Forgot to add this.

Nick Young
– Nick Young

2022-03-27 06:45:48 +00:00
Commented Mar 27, 2022 at 6:45 — Nick Young
– Nick Young, Commented Mar 27, 2022 at 6:45

Hamzah Al-Qadasi · Accepted Answer · 2022-03-27 10:23:24Z

1

I will try to do it with small example:

input1 = torch.full((3, 3), 1)
input2 = torch.full((3, 3), 2)
input3 = torch.full((3, 3), 3)

out = torch.concat((input1,input2, input3)).T.flatten()
torch.stack(torch.split(out, 3), dim=1).reshape(3,-1)

#output

tensor([[1, 2, 3, 1, 2, 3, 1, 2, 3],
        [1, 2, 3, 1, 2, 3, 1, 2, 3],
        [1, 2, 3, 1, 2, 3, 1, 2, 3]])

answered Mar 27, 2022 at 10:23

Hamzah Al-Qadasi

10k3 gold badges29 silver badges54 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

How to alternatively concatenate pytorch tensors?

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related