tile() in PyTorch

By coursity / June 23, 2024

tile() can repeat the zero or more elements of a 0D or more D tensor as shown below:

*Memos:

tile() can be used with torch or a tensor.
The 1st argument(tensor of int, float, complex or bool) with torch or using a tensor(tensor of int, float, complex or bool) is input(Required).
The 2nd argument(tuple) with torch or the 1st argument(tuple) with a tensor is dims(Required). *Memos:
- If at least one dimension is 0, an empty tensor is returned.
- The 1st argument…(int, int…) with a tensor is also dims. *dims= mustn’t be used.

import torch my_tensor = torch.tensor([3, 5, 1]) torch.tile(input=my_tensor, dims=(1,))
my_tensor.tile(dims=(1,))
# tensor([3, 5, 1])

torch.tile(input=my_tensor, dims=(2,))
# tensor([3, 5, 1, 3, 5, 1])

torch.tile(input=my_tensor, dims=(3,))
# tensor([3, 5, 1, 3, 5, 1, 3, 5, 1])
etc. torch.tile(input=my_tensor, dims=(1, 1))
# tensor([[3, 5, 1]])

torch.tile(input=my_tensor, dims=(1, 2))
# tensor([[3, 5, 1, 3, 5, 1]])

torch.tile(input=my_tensor, dims=(1, 3))
# tensor([[3, 5, 1, 3, 5, 1, 3, 5, 1]])
etc. torch.tile(input=my_tensor, dims=(2, 1))
# tensor([[3, 5, 1],
# [3, 5, 1]])

torch.tile(input=my_tensor, dims=(2, 2))
# tensor([[3, 5, 1, 3, 5, 1],
# [3, 5, 1, 3, 5, 1]])

torch.tile(input=my_tensor, dims=(2, 3))
# tensor([[3, 5, 1, 3, 5, 1, 3, 5, 1],
# [3, 5, 1, 3, 5, 1, 3, 5, 1]])
etc. torch.tile(input=my_tensor, dims=(3, 1))
# tensor([[3, 5, 1],
# [3, 5, 1],
# [3, 5, 1]])
etc. torch.tile(input=my_tensor, dims=(1, 1, 1))
# tensor([[[3, 5, 1]]])
etc. torch.tile(input=my_tensor, dims=(1, 0, 1))
# tensor([], size=(1, 0, 3), dtype=torch.int64)

my_tensor.tile(3, 2, 1)
# tensor([[[3, 5, 1], [3, 5, 1]],
# [[3, 5, 1], [3, 5, 1]],
# [[3, 5, 1], [3, 5, 1]]])

my_tensor = torch.tensor([3., 5., 1.]) torch.tile(input=my_tensor, dims=(2,))
# tensor([3., 5., 1., 3., 5., 1.])

my_tensor = torch.tensor([3.+0.j, 5.+0.j, 1.+0.j]) torch.tile(input=my_tensor, dims=(2,))
# tensor([3.+0.j, 5.+0.j, 1.+0.j, 3.+0.j, 5.+0.j, 1.+0.j])

my_tensor = torch.tensor([True, False, True]) torch.tile(input=my_tensor, dims=(2,))
# tensor([True, False, True, True, False, True])

my_tensor = torch.tensor([[3, 5, 1], [6, 0, 5]])
torch.tile(input=my_tensor, dims=(1,))
# tensor([[3, 5, 1],
# [6, 0, 5]])

torch.tile(input=my_tensor, dims=(2,))
# tensor([[3, 5, 1, 3, 5, 1],
# [6, 0, 5, 6, 0, 5]])

torch.tile(input=my_tensor, dims=(3,))
# tensor([[3, 5, 1, 3, 5, 1, 3, 5, 1],
# [6, 0, 5, 6, 0, 5, 6, 0, 5]])
etc. torch.tile(input=my_tensor, dims=(1, 1))
# tensor([[3, 5, 1],
# [6, 0, 5]])

torch.tile(input=my_tensor, dims=(1, 2))
# tensor([[3, 5, 1, 3, 5, 1],
# [6, 0, 5, 6, 0, 5]])

torch.tile(input=my_tensor, dims=(1, 3))
# tensor([[3, 5, 1, 3, 5, 1, 3, 5, 1],
# [6, 0, 5, 6, 0, 5, 6, 0, 5]])
etc. torch.tile(input=my_tensor, dims=(2, 1))
# tensor([[3, 5, 1],
# [6, 0, 5],
# [3, 5, 1],
# [6, 0, 5]])

torch.tile(input=my_tensor, dims=(2, 2))
# tensor([[3, 5, 1, 3, 5, 1],
# [6, 0, 5, 6, 0, 5],
# [3, 5, 1, 3, 5, 1],
# [6, 0, 5, 6, 0, 5]])

torch.tile(input=my_tensor, dims=(2, 3))
# tensor([[3, 5, 1, 3, 5, 1, 3, 5, 1],
# [6, 0, 5, 6, 0, 5, 6, 0, 5],
# [3, 5, 1, 3, 5, 1, 3, 5, 1],
# [6, 0, 5, 6, 0, 5, 6, 0, 5]])
etc. torch.tile(input=my_tensor, dims=(3, 1))
# tensor([[3, 5, 1],
# [6, 0, 5],
# [3, 5, 1],
# [6, 0, 5],
# [3, 5, 1],
# [6, 0, 5]])
etc. torch.tile(input=my_tensor, dims=(1, 1, 1))
# tensor([[[3, 5, 1],
# [6, 0, 5]]])
etc.

Enter fullscreen mode
Exit fullscreen mode

Related

Discover more from Coursity

Subscribe to get the latest posts sent to your email.

Leave a Comment Cancel Reply

Blockchain powered autonomous vehicles : securing data in vehicle to everything (v2x) communication. Copyright © 2024 solid future. There are different types of web hosting services available, such.