heaviside() and Identity() in PyTorch

#pytorch #heaviside #identity #activationfunction

*Memos:

My post explains Step function, Identity and ReLU.
My post explains ReLU() and LeakyReLU().
My post explains PReLU() and ELU().
My post explains SELU() and CELU().
My post explains GELU() and Mish().
My post explains SiLU() and Softplus().
My post explains Tanh() and Softsign().
My post explains Sigmoid() and Softmax().

heaviside() can get the 0D or more D tensor of the zero or more values computed by Heaviside step function from the 0D or more D tensor of zero or more elements as shown below:

*Memos:

heaviside() can be used with torch or a tensor.
The 1st argument(input) with torch or using a tensor(Required-Type:tensor of int, float or bool).
The 2nd argument with torch or the 1st argument with a tensor is values(Required-Type:tensor of int, float or bool).

import torch
from torch import nn

my_tensor = torch.tensor([8, -3, 0, 1, 5, -2, -1, 4])

torch.heaviside(input=my_tensor,
                values=torch.tensor(0))
my_tensor.heaviside(values=torch.tensor(0))
# tensor([1, 0, 0, 1, 1, 0, 0, 1])

torch.heaviside(input=my_tensor,
                values=torch.tensor([0, 1, 2, 3, 4, 5, 6, 7]))
# tensor([1, 0, 2, 1, 1, 0, 0, 1])

my_tensor = torch.tensor([[8, -3, 0, 1],
                          [5, 0, -1, 4]])
torch.heaviside(input=my_tensor, values=torch.tensor(0))
# tensor([[1, 0, 0, 1],
#         [1, 0, 0, 1]])

torch.heaviside(input=my_tensor,
                values=torch.tensor([[0, 1, 2, 3],
                                     [4, 5, 6, 7]]))
# tensor([[1, 0, 2, 1],
#         [1, 5, 0, 1]])

my_tensor = torch.tensor([[[8, -3], [0, 1]],
                          [[5, 0], [-1, 4]]])
torch.heaviside(input=my_tensor, values=torch.tensor(0))
# tensor([[[1, 0], [0, 1]],
#         [[1, 0], [0, 1]]])

torch.heaviside(input=my_tensor,
                values=torch.tensor([[[0, 1], [2, 3]],
                                     [[4, 5], [6, 7]]]))
# tensor([[[1, 0], [2, 1]],
#         [[1, 5], [0, 1]]])

my_tensor = torch.tensor([[[8., -3.], [0., 1.]],
                          [[5., 0.], [-1., 4.]]])
torch.heaviside(input=my_tensor,
                values=torch.tensor([[[0., 1.], [2., 3.]],
                                      [[4., 5.], [6., 7.]]]))
# tensor([[[1., 0.], [2., 1.]],
#         [[1., 5.], [0., 1.]]])

my_tensor = torch.tensor([[[True, False], [True, False]],
                          [[False, True], [False, True]]])
torch.heaviside(input=my_tensor,
                values=torch.tensor([[[True, False], [True, False]],
                                     [[False, True], [False, True]]]))
# tensor([[[True, False], [True, False]],
#         [[False, True], [False, True]]])

Identity() can just get the same tensor as the input tensor which is the 0D or more D tensor of zero or more elements as shown below:
*Memos:

For initialization, you can set 0 or more arguments but there is no influence.
The 1st argument is input(Required-Type:tensor of int or float).

import torch
from torch import nn

my_tensor = torch.tensor([8, -3, 0, 1, 5, -2, -1, 4])

identity = nn.Identity()
identity(input=my_tensor)
# tensor([8, -3, 0, 1, 5, -2, -1, 4])

identity
# Identity()

identity = nn.Identity(num1=3, num2=5)
identity(input=my_tensor)
# tensor([8, -3, 0, 1, 5, -2, -1, 4])

my_tensor = torch.tensor([[8, -3, 0, 1],
                          [5, -2, -1, 4]])
identity = nn.Identity()
identity(input=my_tensor)
# tensor([[8, -3, 0, 1],
#         [5, -2, -1, 4]])

my_tensor = torch.tensor([[[8, -3], [0, 1]],
                          [[5, -2], [-1, 4]]])
identity = nn.Identity()
identity(input=my_tensor)
# tensor([[[8, -3], [0, 1]],
#         [[5, -2], [-1, 4]]])

my_tensor = torch.tensor([[[8., -3.], [0., 1.]],
                          [[5., -2.], [-1., 4.]]])
identity = nn.Identity()
identity(input=my_tensor)
# tensor([[[8., -3.], [0., 1.]],
#         [[5., -2.], [-1., 4.]]])

DEV Community

heaviside() and Identity() in PyTorch

Top comments (0)

Read next

Building a Traffic Grid Analysis with Python: Reflecting on Growth and Tackling Challenges

Installing Kubernetes via MicroK8s and configuring the deployment of NestJS and Angular applications

AWS Observability Maturity Model - V2

AI-powered Github leaderboard.