oneflow.nn.Softplus

class oneflow.nn.Softplus(beta: int = 1, threshold: int = 20)

Applies the element-wise function:

\[\text{Softplus}(x) = \frac{1}{\beta} * \log(1 + \exp(\beta * x))\]

SoftPlus is a smooth approximation to the ReLU function and can be used to constrain the output of a machine to always be positive.

For numerical stability the implementation reverts to the linear function when \(input \times \beta > threshold\).

Parameters

• beta – the \(\beta\) value for the Softplus formulation. Default: 1

• threshold – values above this revert to a linear function. Default: 20

Shape:

• Input: \((N, *)\) where * means, any number of additional dimensions

• Output: \((N, *)\), same shape as the input

For example:

>>> import numpy as np
>>> import oneflow as flow

>>> x = np.array([-0.5, 0, 0.5]).astype(np.float32)
>>> input = flow.Tensor(x)
>>> softplus = flow.nn.Softplus()

>>> out = softplus(input)
>>> out
tensor([0.4741, 0.6931, 0.9741], dtype=oneflow.float32)