split in PyTorch

#python #pytorch #split #function

*Memos:

My post explains vsplit().
My post explains hsplit().
My post explains dsplit().
My post explains tensor_split().
My post explains chunk().
My post explains unbind().

split() can get the one or more 1D or more D splitted view tensors of zero or more elements from the 1D or more D tensor of zero or more elements as shown below:

*Memos:

split() can be used with torch or a tensor.
The 1st argument(tensor) with torch or using a tensor(Required-Type:tensor of int, float, complex or bool).
The 2nd argument with torch or the 1st argument with a tensor is split_size_or_sections(Required-Type:int, tuple of int or list of int). *Don't use split_size_or_sections= with a tensor.
The 3rd argument with torch or the 2nd argument with a tensor is dim(Optional-Default:0-Type:int).
The total number of the zero or more elements of one or more returned tensors changes.
One or more returned tensors keep the dimension of the original tensor.

import torch

my_tensor = torch.tensor([[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]])

torch.split(tensor=my_tensor, split_size_or_sections=1)
my_tensor.split(1)
torch.split(tensor=my_tensor, split_size_or_sections=1, dim=0)
torch.split(tensor=my_tensor, split_size_or_sections=1, dim=-2)
# (tensor([[0, 1, 2, 3]]),
#  tensor([[4, 5, 6, 7]]),
#  tensor([[8, 9, 10, 11]]))

torch.split(tensor=my_tensor, split_size_or_sections=1, dim=1)
torch.split(tensor=my_tensor, split_size_or_sections=1, dim=-1)
# (tensor([[0], [4], [8]]),
#  tensor([[1], [5], [9]]),
#  tensor([[2], [6], [10]]),
#  tensor([[3], [7], [11]]))

torch.split(tensor=my_tensor, split_size_or_sections=2)
torch.split(tensor=my_tensor, split_size_or_sections=2, dim=0)
torch.split(tensor=my_tensor, split_size_or_sections=2, dim=-2)
# (tensor([[0, 1, 2, 3], [4, 5, 6, 7]]),
#  tensor([[8, 9, 10, 11]]))

torch.split(tensor=my_tensor, split_size_or_sections=2, dim=1)
torch.split(tensor=my_tensor, split_size_or_sections=2, dim=-1)
# (tensor([[0, 1], [4, 5], [8, 9]]),
#  tensor([[2, 3], [6, 7], [10, 11]]))

torch.split(tensor=my_tensor, split_size_or_sections=3)
torch.split(tensor=my_tensor, split_size_or_sections=3, dim=0)
torch.split(tensor=my_tensor, split_size_or_sections=3, dim=-2)
# (tensor([[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]),)

torch.split(tensor=my_tensor, split_size_or_sections=3, dim=1)
torch.split(tensor=my_tensor, split_size_or_sections=3, dim=-1)
# (tensor([[0, 1, 2], [4, 5, 6], [8, 9, 10]]),
#  tensor([[3], [7], [11]]))

torch.split(tensor=my_tensor, split_size_or_sections=(0, 3))
torch.split(tensor=my_tensor, split_size_or_sections=(0, 3), dim=0)
torch.split(tensor=my_tensor, split_size_or_sections=(0, 3), dim=-2)
# (tensor([], size=(0, 4), dtype=torch.int64),
#  tensor([[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]))

torch.split(tensor=my_tensor, split_size_or_sections=(1, 2))
torch.split(tensor=my_tensor, split_size_or_sections=(1, 2), dim=0)
torch.split(tensor=my_tensor, split_size_or_sections=(1, 2), dim=-2)
# (tensor([[0, 1, 2, 3]]),
#  tensor([[4, 5, 6, 7], [8, 9, 10, 11]]))

torch.split(tensor=my_tensor, split_size_or_sections=(2, 1))
torch.split(tensor=my_tensor, split_size_or_sections=(2, 1), dim=0)
torch.split(tensor=my_tensor, split_size_or_sections=(2, 1), dim=-2)
# (tensor([[0, 1, 2, 3], [4, 5, 6, 7]]),
#  tensor([[8, 9, 10, 11]]))

torch.split(tensor=my_tensor, split_size_or_sections=(3, 0))
torch.split(tensor=my_tensor, split_size_or_sections=(3, 0), dim=0)
# (tensor([[0, 1, 2, 3], [4, 5, 6, 7], [8, 9, 10, 11]]),
#  tensor([], size=(0, 4), dtype=torch.int64))

torch.split(tensor=my_tensor, split_size_or_sections=(1, 1, 1))
torch.split(tensor=my_tensor, split_size_or_sections=(1, 1, 1), dim=0)
torch.split(tensor=my_tensor, split_size_or_sections=(1, 1, 1), dim=-2)
# (tensor([[0, 1, 2, 3]]),
#  tensor([[4, 5, 6, 7]]), 
#  tensor([[ 8,  9, 10, 11]]))

my_tensor = torch.tensor([[0., 1., 2., 3.],
                          [4., 5., 6., 7.],
                          [8., 9., 10., 11.]])
torch.split(tensor=my_tensor, split_size_or_sections=1)
# (tensor([[0., 1., 2., 3.]]),
#  tensor([[4., 5., 6., 7.]]),
#  tensor([[8., 9., 10., 11.]]))

my_tensor = torch.tensor([[0.+0.j, 1.+0.j, 2.+0.j, 3.+0.j],
                          [4.+0.j, 5.+0.j, 6.+0.j, 7.+0.j],
                          [8.+0.j, 9.+0.j, 10.+0.j, 11.+0.j]])
torch.split(tensor=my_tensor, split_size_or_sections=1)
# (tensor([[0.+0.j, 1.+0.j, 2.+0.j, 3.+0.j]]),
#  tensor([[4.+0.j, 5.+0.j, 6.+0.j, 7.+0.j]]),
#  tensor([[8.+0.j, 9.+0.j, 10.+0.j, 11.+0.j]]))

my_tensor = torch.tensor([[True, False, True, False],
                          [False, True, False, True],
                          [True, False, True, False]])
torch.split(tensor=my_tensor, split_size_or_sections=1)
# (tensor([[True, False, True, False]]),
#  tensor([[False, True, False, True]]),
#  tensor([[True, False, True, False]]))

DEV Community

split in PyTorch

Top comments (0)

Read next

FireDucks: Get performance beyond pandas with zero learning cost!

Investigating the performance of np.einsum

How to Host a Laravel Project in a Subdirectory on Shared Hosting without Exposing `/public` in the URL

Introduction to Django's Default User Model