struct

`SushiAI::NN::LeakyReLU`

Applies x where x is positive and slope * x elsewhere, elementwise.

Public attributes

double slope = 0.01

Holds the factor applied to negative inputs; must lie in (0, 1).

static constexpr auto fields()

Lists this layer's parameters, of which there are none.

Graph::ValueId forward(const Graph::GraphBuilder &builder, Graph::ValueId x) const

Traces the activation over an input of any rank.

Returns

The result's id, shaped and laid out like x.

Shape output_shape(const Shape &input) const

Computes the per-sample shape forward() produces.

Returns

The shape of input, which the activation keeps.