How to Create a Boolean Tensor in PyTorch

To create a Boolean tensor efficiently in PyTorch, use the torch.tensor() method and pass the dtype=torch.bool explicitly (inferred if data is bools), where each element holds a True or False value.

import torch

bool_tensor = torch.tensor([True, False, True], dtype=torch.bool)

print(bool_tensor)

# Output: tensor([ True, False,  True], dtype=torch.bool)

You can use this boolean tensor for masking, condition checks, logical operations, or indexing.

Tensor initialization methods

Generating All False using torch.zeros()

If you want to create a Boolean tensor where each value is False, you can use torch.zeros() method.

import torch

bool_false = torch.zeros(3, dtype=torch.bool)

print(bool_false)

# Output: tensor([False, False, False])

In this code, we created a Boolean tensor with three elements, all of which are False.

Generating All True using torch.ones()

If you want to create a Boolean tensor where each value is True, you can use torch.ones() method.

import torch

bool_true = torch.ones(3, dtype=torch.bool)

print(bool_true)

# Output: tensor([ True,  True,  True])

In this code, we created a Boolean tensor with three elements, all of which are True.

Fill with a value using torch.full()

Using torch.full() method, you can create a Boolean tensor of either True values or False.

import torch

bool_full_true = torch.full((4,), True, dtype=torch.bool)

print(bool_full_true)

# Output: tensor([True, True, True, True])

bool_full_false = torch.full((4,), False, dtype=torch.bool)

print(bool_full_false)

# Output: tensor([False, False, False, False])

In this code, we created both True and False-valued tensors.

That’s all!