Module activation_function 

Neural network activation functions with forward pass, derivative, and vectorized operations.

Provides a comprehensive set of activation functions for neural network layers including ReLU variants, sigmoid, tanh, softmax, GELU, Swish, Mish, HardSwish, and more. Each activation supports forward evaluation, derivative computation for backpropagation, in-place mutation, and call statistics tracking.

Naming Convention

Types in this module are prefixed with Af (Activation Function) in re-exports to avoid collision with the pre-existing activation module in the same crate.

Structs

ActivationConfig

Configuration for an ActivationFunction instance.

ActivationFunction

Neural network activation layer with forward pass, derivative, in-place application, and call statistics.

ActivationStats

Runtime statistics collected by an ActivationFunction.

Enums

ActivationType

Specifies which activation function to apply.