PyTorch Stack Tutorial + Examples

The PyTorch torch.stack() function is used to concatenate the tensor with the same dimension and shape. In detail, we will discuss the stack() function using PyTorch in Python.

And additionally, we will cover different examples related to the PyTorch stack function. And we will cover these topics.

• What is PyTorch stack
• PyTorch stack example
• How to use Pytorch stack tensors
• PyTorch 2d tensor to be stack to create a 3d tensor
• PyTorch 1d tensor stacked and generate a 2d tensor as the final tensor
• PyTorch stack to show error when the input tensor are not of the same shape

What is PyTorch stack

In this section, we will learn about the PyTorch stack in python.

The PyTorch stack() method is used to join or concatenate a series of a tensor along with a new dimension. This function is used to concatenate the tensor with the same dimension and shape.

Syntax:

Syntax of the PyTorch stack:

torch.stack(tensors, dim=0, out=None)

Parameters:

The following are the parameters of the PyTorch stack:

• tensors: The tensor is defined as a series of tensors along with new dimensions.
• dim: The dim is the dimension to insert and its integer lies between 0 and the number of dimensions of input tensors.
• out: The out is a parameter that describes the output tensor.

This is how we understand about the Pytorch stack with the help of a torch.stack() function.

Read: How to use PyTorch Full() Function

PyTorch stack example

In this section, we will learn how we implement the PyTorch stack with the help of an example in python.

The torch.stack() method in which all the tensors need to be of the same size and used to join or concatenate a series of a tensor along with a new dimension.

Code:

In the following code, we will import the torch module such as import torch.

• stackt1 = torch.tensor([1.,3.,6.,10.]): Here we are creating a tensor by using torch.tensor() function.
• print(“Tensor stackt1:”, stackt1) is used to print the created tensor by using print() function.
• s = torch.stack((stackt1,stackt2)): Here we are calling the stack() function.
• print(s): Here we print the final tensor after joining with the help of the print() function.

# importing torch
import torch
  
# Creating a tensors
stackt1 = torch.tensor([1.,3.,6.,10.])
stackt2 = torch.tensor([2.,7.,9.,13.])
  
# printing above created tensors
print("Tensor stackt1:", stackt1)
print("Tensor ystackt2:", stackt2)
  
# join above tensor using "torch.stack()"
print("Join the tensors:")
s = torch.stack((stackt1,stackt2))
  
# print final tensor after join
print(s)
  
print("Join the tensors dimension 0:")
s = torch.stack((stackt1,stackt2), dim = 0)
print(s)
  
print("join tensors dimension 1:")
s = torch.stack((stackt1,stackt2), dim = 1)
print(s)

Output:

After running the above code, we get the following output in which we can see that the PyTorch stack values are printed on the screen.

So, with this, we understood about the PyTorch stack with the help of an example in python.

Read: Create PyTorch Empty Tensor

How to use Pytorch stack tensors

In this section, we will learn about the PyTorch stack tensor in python.

The PyTorch stack tensor is defined as a method that concatenates a sequence of two or more tensors along a new dimension.

Code:

In the following code, we will import the torch library as import torch.

• stacktensor_1 = torch.Tensor([[20,30,40],[50,60,70]]): Here we are describing the variable by using torch.tensor() function.
• print(“stacktensor_1 \n”, stacktensor_1) is used to print the first tensor by using print() function.
• stacktensor = torch.stack((stacktensor_1, stacktensor_2), -1): Here we are calling the torch.stack() function.
• print(“Join the tensors in the -1 dimension \n”, stacktensor) is used to print the stack tensor value by using the print() function.

# import torch library
import torch
  
# define tensors
stacktensor_1 = torch.Tensor([[20,30,40],[50,60,70]])
stacktensor_2 = torch.Tensor([[40,50,60],[70,80,90]])
  
# print first tensors
print("stacktensor_1 \n", stacktensor_1)
  
# print second tensor
print("stacktensor_2 \n", stacktensor_2)
  
# Calling the torch.stack() function join tensor in -1 dimension
stacktensor = torch.stack((stacktensor_1, stacktensor_2), -1)
print("Join the tensors in the -1 dimension \n", stacktensor)
  
# Join the tensor in 0 dimension
stacktensor = torch.stack((stacktensor_1, stacktensor_2), 0)
print("Join the tensors in the 0 dimension \n", stacktensor)

Output:

After running the above code, we get the following output in which we can see that the PyTorch stack tensor values are printed on the screen.

This is how we understand about the PyTorch stack tensor by using a torch.stack() function.

Read: Introduction to PyTorch Lenet

PyTorch 2d tensor to be stack to create a 3d tensor

In this section, we will learn about the PyTorch 2d tensor to be stack to create a 3d tensor in python.

Here we are using the two-dimensional tensor and we joined all the tensors together and create the three-dimensional tensor.

Code:

In the following code, we will import the torch library such as import torch.

• stacktens1 = torch.Tensor([[2, 4], [4, 8]]): Here we are declaring the variable by using torch.tensor() function.
• print(“\n Stack First Tensor :\n”, stacktens1) is used to print the stack first tensor by using print() function.
• stacktens = torch.stack((stacktens1, stacktens2, stacktens3), -1): Here we are stacked tensor in -1 dimension.
• print(“\n tensors in -1 dimension \n”, stacktens) is used to print the tensor in -1 dimension with the help of the print() function.
• stacktens = torch.stack((stacktens1, stacktens2, stacktens3), 0): Here we are stacked the tensors in 0 dimension.

# import library
import torch
  
# declaring some tensors
stacktens1 = torch.Tensor([[2, 4], [4, 8]])
stacktens2 = torch.Tensor([[3, 4], [5, 10]])
stacktens3 = torch.Tensor([[8, 16], [9, 18]])
  
# Display the tensors
print("\n Stack First Tensor :\n", stacktens1)
print("\n Stack Second Tensor :\n", stacktens2)
print("\n Stack Third Tensor :\n", stacktens3)
  
# stacked tensors in -1 dimension
stacktens = torch.stack((stacktens1, stacktens2, stacktens3), -1)
print("\n tensors in -1 dimension \n", stacktens)
  
# stacked tensors in 0 dimension
stacktens = torch.stack((stacktens1, stacktens2, stacktens3), 0)
print("\n tensors in 0 dimension \n", stacktens)

Output:

In the below output you can see that the PyTorch 2d tensor is to be stacked to create 3d tensor values that are printed on the screen.

So, with this, we understood about the PyTorch 2d tensor to be stacked to create a 3d tensor.

Read: PyTorch Add Dimension

PyTorch 1d tensor stacked and generate a 2d tensor as final tensor

In this section, we will learn about the PyTorch 1d tensor stacked and generate a 2d tensor as the final tensor in python.

Here we are generating the two-dimensional tensor as a final tensor from the PyTorch one-dimensional tensor by using the torch.stack() function.

Code:

In the following code, we will import the required library such as import torch.

• stacktens1 = torch.Tensor([2, 4, 6]): Here we are defining the tensor by using the torch.tensor() function.
• print(” Stack First Tensor :\n”, stacktens1) is used to print the stack first tensor with the help of the print() function.
• stacktens = torch.stack((stacktens1, stacktens2, stacktens3), 0): Here we are calling the stack() function.

# import required library
import torch
  
# Define some tensors
stacktens1 = torch.Tensor([2, 4, 6])
stacktens2 = torch.Tensor([3, 6, 9])
stacktens3 = torch.Tensor([4, 8, 12])
  
# Display the tensors
print(" Stack First Tensor :\n", stacktens1)
print("\nStack Second Tensor :\n", stacktens2)
print("\n Stack Third Tensor :\n", stacktens3)
  
# Stack the tensors in the 0 dimension
stacktens = torch.stack((stacktens1, stacktens2, stacktens3), 0)
print("\n Join the tensors in the 0 dimension \n", stacktens)
  
#  Stack the tensors in the -1 dimension
stacktens = torch.stack((stacktens1, stacktens2, stacktens3), -1)
print("\n Join the tensors in the -1 dimension \n", stacktens)

Parameters:

After running the above code, we get the following output in which we can see that the PyTorch 1d tensor is stacked and generate a 2d tensor as the final tensor and the values are printed on the screen.

This is how we understand how the PyTorch 1d tensor is stacked and generate a 2d tensor as the final tensor.

Read: PyTorch nn Conv2d

PyTorch stack to show error when the input tensor are not of the same shape

In this section, we will learn about the PyTorch stack to show error when the input tensor is not of the same shape in python.

Here we are using the PyTorch stack that can show the error when the input tensor does not have a similar shape.

Code:

In the following code, we will import the necessary library such as import torch.

• stacktens1 = torch.tensor([1., 3., 6., 10.]): Here we are creating the stack tensor by using a torch.tensor() function.
• print(“Tensor stacktens:”, stacktens1) is used to print the tensor stack by using the print() function.
• s = torch.stack((stacktens1, stacktens2)): Here we are calling the stack() fucntion.

# importing torch
import torch
  
# Creating the stack tensors
stacktens1 = torch.tensor([1., 3., 6., 10.])
stacktens2 = torch.tensor([2., 7., 9.])
  
# Printing the tensors
print("Tensor stacktens:", stacktens1)
print("Tensor stac:", stacktens2)
  
# Stack the tensor using "torch.stack()"
print("Stack the tensors:")
s = torch.stack((stacktens1, stacktens2))
# Print the final tensor after join
print(s)
  
print("Stack the tensors dimension 0:")
s = torch.stack((stacktens1, stacktens2), dim=0)
print(s)
  
print("Stack the tensors dimension 1:")
s = torch.stack((stacktens1, stacktens2), dim=1)
print(s)

Output:

After running the above code we get the following output in which we can see that the error is shown when the input tensor is not of the same shape.

This is how we understand how the Pytorch stack show error when the input tensor is not of the same shape.

Also, take a look at some more PyTorch tutorials.

• PyTorch Leaky ReLU
• PyTorch Cat function
• PyTorch Activation Function
• PyTorch RNN – Detailed Guide
• PyTorch MNIST Tutorial

So, in this tutorial, we discussed PyTorch Stack and we have also covered different examples related to its implementation. Here is the list of examples that we have covered.

• What is PyTorch stack?
• PyTorch stack example
• How to use Pytorch stack tensors
• PyTorch 2d tensor to be stack to create a 3d tensor
• PyTorch 1d tensor stacked and generate a 2d tensor as the final tensor
• PyTorch stack to show error when the input tensor are not of the same shape