Enum RnnActivation

Options for activation functions to apply in a recurrent layer.

Namespace: Unity.Sentis.Layers

Assembly: solution.dll

Syntax

public enum RnnActivation

Fields

Name	Description
Affine	Use Affine activation: f(x) = alpha * x + beta.
Elu	Use Elu activation: f(x) = x if x >= 0, otherwise f(x) = alpha * (e^x - 1).
HardSigmoid	Use HardSigmoid activation: f(x) = clamp(alpha * x + beta, 0, 1).
LeakyRelu	Use LeakyRelu activation: f(x) = x if x >= 0, otherwise f(x) = alpha * x.
Relu	Use Relu activation: f(x) = max(0, x).
ScaledTanh	Use ScaledTanh activation: f(x) = alpha * tanh(beta * x).
Sigmoid	Use Sigmoid activation: f(x) = 1 / (1 + e^{-x}).
Softplus	Use Softplus activation: f(x) = log(1 + e^x).
Softsign	Use Softsign activation: f(x) = x / (1 + |x|).
Tanh	Use Tanh activation: f(x) = (1 - e^{-2x}) / (1 + e^{-2x}).
ThresholdedRelu	Use ThresholdedRelu activation: f(x) = x if x >= alpha, otherwise f(x) = 0.