torch Tensor add dimension

Question

I have a tensor with this size

torch.Size([128, 64])

how do I add one "dummy" dimension such as

torch.Size([1, 128, 64])

torch.unsqueeze

KDecker
– KDecker

2021-07-20 13:07:36 +00:00
Commented Jul 20, 2021 at 13:07 — KDecker
– KDecker, Commented Jul 20, 2021 at 13:07
using torch.unsqueeze(0)along the first dimension.

Priya
– Priya

2021-07-20 13:28:47 +00:00
Commented Jul 20, 2021 at 13:28 — Priya
– Priya, Commented Jul 20, 2021 at 13:28

Shai · Accepted Answer · 2021-07-20 13:29:04Z

3

There are several ways:

torch.unsqueeze:

torch.unsqueeze(x, 0)

Using None (or np.newaxis):

x[None, ...]
# or
x[np.newaxis, ...]

reshape or view:

x.reshape(1, *x.shape)
# or
x.view(1, *x.shape)

answered Jul 20, 2021 at 13:29

Shai

115k39 gold badges259 silver badges398 bronze badges

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

3 Comments

Ivan Over a year ago

Or x[None] ;)

George Ogden Over a year ago

Do not use x[None] or x[None,...]! Use x[np.newaxis] or x[np.newaxis, ...] ("Explicit is better than implicit.")

Shai Over a year ago

@GeorgeOgden you are right about np.newaxis. Nevertheless, I like it better to use just None. Moreover, in some cases, I do not need/want to explicitly import numpy in my code.

Collectives™ on Stack Overflow

torch Tensor add dimension

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related