reference/generated/tensordict.nn.biased_softplus

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

biased_softplus

class tensordict.nn.biased_softplus(bias: float, min_val: float = 0.01)

Bases: Module

A softplus shifted so that it returns bias for a zero input and stays above min_val.

The module computes softplus(x + inv_softplus(bias - min_val)) + min_val. mappings() returns it for the keys "biased_softplus_{bias}" and "biased_softplus_{bias}_{min_val}".

Parameters:

• bias (float) – the output for a zero input.

• min_val (float, optional) – the lower bound of the output. Defaults to 0.01.

Examples

>>> import torch
>>> from tensordict.nn import biased_softplus
>>> f = biased_softplus(1.0)
>>> f(torch.zeros(2))
tensor([1., 1.])

forward(x: Tensor) → Tensor

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.