In this section we are going to get introduced to the basics of PyTorch. We will learn how to create tensors, perform basic operations on them, and convert them to NumPy arrays and back.
First, we need to install PyTorch. You can do this using pip:
pip install torch
Next, we will import the PyTorch module in our Python script and make sure it is working:
Tensors are a generalized version of matrices and are the fundamental building blocks of PyTorch.
To create a tensor in PyTorch, you can use the torch.Tensor() function:
import torch # Creating a 1D Tensor a = torch.Tensor([1,2,3]) print(a) # Creating a 2D Tensor b = torch.Tensor([[1,2,3],[4,5,6]]) print(b)
Tensors have attributes like shape, dtype, and device which they are stored in.
# Shape of a tensor print(b.shape) # or b.size() # Data type of a tensor print(b.dtype) # Device the tensor is stored on print(b.device)
We can perform various operations on tensors.
# Addition c = torch.Tensor([1,2,3]) d = torch.Tensor([4,5,6]) print(c+d) # Multiplication (Element-wise) print(c*d) # Matrix Multiplication e = torch.Tensor([[1,2],[3,4]]) f = torch.Tensor([[5,6],[7,8]]) print(e.matmul(f)) # or torch.mm(e, f)
You can change the shape of a tensor without changing its data using the view() method.
g = torch.Tensor([1,2,3,4,5,6]) print(g.view(2,3)) # Reshapes g to a 2x3 matrix
We can easily convert a NumPy array to a PyTorch tensor and vice versa.
import numpy as np # Creating a NumPy array numpy_array = np.array([1, 2, 3, 4, 5]) # Converting the NumPy array to a PyTorch tensor tensor_from_numpy = torch.from_numpy(numpy_array) print(tensor_from_numpy) # Converting a PyTorch tensor to a NumPy array numpy_from_tensor = tensor_from_numpy.numpy() print(numpy_from_tensor)
These basics will set the foundation for your learning of more advanced PyTorch concepts and operations.