5 Useful Tensor Operations in PyTorch

This notebook discusses about 5 useful tensor operations that help in matrix manipulation, conditional operation and reshaping techniques.

PyTorch is an open source machine learning library based on the Torch library, used for applications such as computer vision and natural language processing, primarily developed by Facebook's AI Research lab.

The functions discussed in this notebook are:

torch.narrow
torch.matmul
torch.where
torch.all
torch.squeeze

# Import torch and other required modules
import torch

Function 1 - torch.narrow

The narrow method returns a narrowed version of the original tensor i.e., used to slice the tensors by defining the dimension, start and length parameters.

m = torch.randn(2,3,4)
print(m)

# torch.narrow(tensor, dimension, start, length)
torch.narrow(m,1,2,1)

tensor([[[ 1.4695, -0.0572, -0.8993,  1.0886],
         [-0.1281,  1.5652, -0.0577,  0.5780],
         [ 0.3634, -1.1730, -0.6827,  0.5164]],

        [[-0.8064, -0.4728, -0.3439, -1.7098],
         [ 0.8269,  0.6465, -1.5112,  0.9079],
         [ 0.5152, -0.3883, -1.4499,  0.6858]]])

tensor([[[ 0.3634, -1.1730, -0.6827,  0.5164]],

        [[ 0.5152, -0.3883, -1.4499,  0.6858]]])

The tensor m is narrowed (sliced) along the dimension(axis) = 1 starting from the index position start = 2 to ending start+length = 2+1 = 3, with ending index exclusive. One important thing to notice is that the dimension value ranges from -3 to 2, both inclusive.