PyTorch .all()

The .all() function returns True if all elements in the input tensor evaluate to True, otherwise it returns False. If a dimension is specified, the function performs a logical AND reduction along that dimension.

Non-zero numeric values evaluate as True, and zero evaluates as False.

Syntax

torch.all(input, dim, keepdim=False, *, out=None)

Or alternatively, in tensor method form:

tensor.all(dim=None, keepdim=False)

Parameters:

• input: The input tensor.
• dim (optional): An int or tuple of int values defining the dimension or dimensions to reduce. If None, all dimensions are reduced.
• keepdim (optional): A bool value that defines if the output tensor retains the reduced dimension. The default value is False.
• out (optional): The output tensor.

Return value:

• If dim is not specified, returns a zero-dimensional boolean tensor containing True or False.
• If dim is specified, returns a tensor reduced along the given dimension(s), containing boolean values.

Example 1: Reducing an Entire Tensor

In this example, torch.all() checks whether all elements in a boolean tensor are True:

import torch

tensor = torch.tensor([True, True, False])
print(torch.all(tensor))

Copy to clipboard
Copy to clipboard

This produces the following output:

tensor(False)

Copy to clipboard
Copy to clipboard

Example 2: Using Numeric Values

In this example, non-zero values evaluate as True and zero evaluates as False:

import torch

tensor = torch.tensor([1, 2, 0])
print(torch.all(tensor))

Copy to clipboard
Copy to clipboard

This produces the following output:

tensor(False)

Copy to clipboard
Copy to clipboard

Codebyte Example: Reducing Along a Dimension

In this example, torch.all() checks values along specific dimensions of a 2D tensor:

import torch

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

print(torch.all(matrix, dim=1))

Copy to clipboard
Copy to clipboard