Softplus#

classtorch.nn.modules.activation.Softplus(beta=1.0,threshold=20.0)[source]#

Applies the Softplus function element-wise.

$$\text{Softplus}(x)=\frac{1}{\beta }\ast \log (1+\exp (\beta \ast x))$$Softplus(x)=β1​∗log(1+exp(β∗x))

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

For numerical stability the implementation reverts to the linear functionwhen$input\times \beta >threshold$input×β>threshold.

Parameters

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

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

Shape:

• Input:$(\ast )$(∗), where$\ast$∗ means any number of dimensions.

• Output:$(\ast )$(∗), same shape as the input.

Examples:

>>>m=nn.Softplus()>>>input=torch.randn(2)>>>output=m(input)

extra_repr()[source]#

Return the extra representation of the module.

Return type

str

forward(input)[source]#

Run forward pass.

Return type

Tensor