Name: repeat_interleave() in PyTorch
Rating: 2.4 (9193 reviews)
Author: hyperkai

*Memos:

My post explains tile().
My post explains repeat().
My post explains expand().

repeat_interleave() can get the 1D tensor of zero or more immediately repeated elements from the 0D or more D tensor of zero or more elements as shown below:

*Memos:

repeat_interleave() can be used with torch or a tensor.
The 1st argument(input) with torch or using a tensor(Optional-Type:tensor of int, float, complex or bool).
The 2nd argument with torch or the 1st argument with a tensor is repeats(Required-Type:int or tensor of int, float, complex or bool). *The tensor must be 0D or 1D.
The 3rd argument with torch or the 2nd argument with a tensor is dim(Optional-Type:int).
There is output_size argument with torch or a tensor(Optional-Default:None-Type:int): *Memos:
- Total output size for the given axis (e.g. sum of repeats). If given, it will avoid stream synchronization needed to calculate output shape of the tensor.
- output_size= must be used.

import torch

my_tensor = torch.tensor([7, 4, 2])

torch.repeat_interleave(repeats=my_tensor)
# tensor([0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2])

torch.repeat_interleave(input=my_tensor, repeats=my_tensor)
my_tensor.repeat_interleave(repeats=my_tensor)
torch.repeat_interleave(input=my_tensor, repeats=my_tensor, dim=0)
torch.repeat_interleave(input=my_tensor, repeats=my_tensor, dim=-1)
# tensor([7, 7, 7, 7, 7, 7, 7, 4, 4, 4, 4, 2, 2])

torch.repeat_interleave(input=my_tensor,
                        repeats=torch.tensor([2, 1, 4]))
torch.repeat_interleave(input=my_tensor,
                        repeats=torch.tensor([2, 1, 4]), dim=0)
torch.repeat_interleave(input=my_tensor,
                        repeats=torch.tensor([2, 1, 4]), dim=-1)
# tensor([7, 7, 4, 2, 2, 2, 2])

torch.repeat_interleave(input=my_tensor, repeats=torch.tensor(2))
torch.repeat_interleave(input=my_tensor, repeats=torch.tensor(2), dim=0)
torch.repeat_interleave(input=my_tensor, repeats=torch.tensor(2), dim=-1)
torch.repeat_interleave(input=my_tensor, repeats=torch.tensor([2]))
torch.repeat_interleave(input=my_tensor, repeats=torch.tensor([2]), dim=0)
torch.repeat_interleave(input=my_tensor, repeats=torch.tensor([2]), dim=-1)
# tensor([7, 7, 4, 4, 2, 2])

torch.repeat_interleave(input=my_tensor, repeats=0)
torch.repeat_interleave(input=my_tensor, repeats=0, dim=0)
torch.repeat_interleave(input=my_tensor, repeats=0, dim=-1)
# tensor([], dtype=torch.int64)

torch.repeat_interleave(input=my_tensor, repeats=1)
torch.repeat_interleave(input=my_tensor, repeats=1, dim=0)
torch.repeat_interleave(input=my_tensor, repeats=1, dim=-1)
# tensor([7, 4, 2])

torch.repeat_interleave(input=my_tensor, repeats=2)
torch.repeat_interleave(input=my_tensor, repeats=2, dim=0)
torch.repeat_interleave(input=my_tensor, repeats=2, dim=-1)
# tensor([7, 7, 4, 4, 2, 2])

torch.repeat_interleave(input=my_tensor, repeats=3)
torch.repeat_interleave(input=my_tensor, repeats=3, dim=0)
torch.repeat_interleave(input=my_tensor, repeats=3, dim=-1)
# tensor([7, 7, 7, 4, 4, 4, 2, 2, 2])
etc.

torch.repeat_interleave(input=my_tensor, repeats=3, dim=0, output_size=9)
# tensor([7, 7, 7, 4, 4, 4, 2, 2, 2])

my_tensor = torch.tensor([[7, 4, 2], [5, 1, 6]])

torch.repeat_interleave(input=my_tensor, repeats=1)
# tensor([7, 4, 2, 5, 1, 6])

torch.repeat_interleave(input=my_tensor, repeats=1, dim=0)
torch.repeat_interleave(input=my_tensor, repeats=1, dim=1)
torch.repeat_interleave(input=my_tensor, repeats=1, dim=-1)
torch.repeat_interleave(input=my_tensor, repeats=1, dim=-2)
# tensor([[7, 4, 2], [5, 1, 6]])

torch.repeat_interleave(input=my_tensor, repeats=2)
# tensor([7, 7, 4, 4, 2, 2, 5, 5, 1, 1, 6, 6])

torch.repeat_interleave(input=my_tensor, repeats=2, dim=0)
torch.repeat_interleave(input=my_tensor, repeats=2, dim=-2)
# tensor([[7, 4, 2], [7, 4, 2], [5, 1, 6], [5, 1, 6]])

torch.repeat_interleave(input=my_tensor, repeats=2, dim=1)
torch.repeat_interleave(input=my_tensor, repeats=2, dim=-1)
# tensor([[7, 7, 4, 4, 2, 2], [5, 5, 1, 1, 6, 6]])

torch.repeat_interleave(input=my_tensor, repeats=3)
# tensor([7, 7, 7, 4, 4, 4, 2, 2, 2, 5, 5, 5, 1, 1, 1, 6, 6, 6])

torch.repeat_interleave(input=my_tensor, repeats=3, dim=0)
torch.repeat_interleave(input=my_tensor, repeats=3, dim=-2)
# tensor([[7, 4, 2], [7, 4, 2], [7, 4, 2], [5, 1, 6], [5, 1, 6], [5, 1, 6]])
torch.repeat_interleave(input=my_tensor, repeats=3, dim=1)
torch.repeat_interleave(input=my_tensor, repeats=3, dim=-1)
# tensor([[7, 7, 7, 4, 4, 4, 2, 2, 2], [5, 5, 5, 1, 1, 1, 6, 6, 6]])

my_tensor = torch.tensor([[7., 4., 2.], [5., 1., 6.]])

torch.repeat_interleave(input=my_tensor, repeats=1)
# tensor([7., 4., 2., 5., 1., 6.])

my_tensor = torch.tensor([[7.+0.j, 4.+0.j, 2.+0.j], [5.+0.j, 1.+0.j, 6.+0.j]])

torch.repeat_interleave(input=my_tensor, repeats=1)
# tensor([7.+0.j, 4.+0.j, 2.+0.j, 5.+0.j, 1.+0.j, 6.+0.j])

my_tensor = torch.tensor([[True, False, True], [False, True, False]])

torch.repeat_interleave(input=my_tensor, repeats=1)
# tensor([True, False, True, False, True, False])