Class LinearLayer

java.lang.Object

smile.deep.layer.ModuleLayer

smile.deep.layer.LinearLayer

All Implemented Interfaces:

AutoCloseable, Function<Tensor,Tensor>, Layer, LinearOp

public class LinearLayer
extends ModuleLayer
implements LinearOp

A fully connected linear layer.

Constructor Summary

Constructors

Constructor	Description
LinearLayer(int in, int out)	Constructor.
LinearLayer(int in, int out, boolean bias)	Constructor.

Method Summary

Modifier and Type	Method	Description
Tensor	forward(Tensor input)	Forward propagation (or forward pass) through the layer.
static LinearLayer	gemm(int in, int out, boolean bias)	A linear layer for a projection that a quantized backend may replace: identical to the constructor, except inside ParameterInit.placeholderGemms() where only a 1x1 placeholder is allocated.
Tensor	weight()	The weight [out, in] as an alias of the module parameter (shares storage; do not close it before the layer is done with).

Methods inherited from class ModuleLayer

close, module, toString

Methods inherited from class Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface Function

andThen, compose

Methods inherited from interface Layer

apply, name, to, to

Constructor Details

LinearLayer

public LinearLayer(int in,
 int out)

Constructor.

Parameters:

in - the number of input features.

out - the number of output features.

LinearLayer

public LinearLayer(int in,
 int out,
 boolean bias)

Constructor.

Parameters:

in - the number of input features.

out - the number of output features.

bias - If false, the layer will not learn an additive bias.

Method Details

gemm

public static LinearLayer gemm(int in,
 int out,
 boolean bias)

A linear layer for a projection that a quantized backend may replace: identical to the constructor, except inside ParameterInit.placeholderGemms() where only a 1x1 placeholder is allocated.

Parameters:

in - the number of input features.

out - the number of output features.

bias - whether the layer has a bias.

Returns:

the layer.

forward

public Tensor forward(Tensor input)

Description copied from interface: Layer

Forward propagation (or forward pass) through the layer.

Specified by:

forward in interface Layer

Specified by:

forward in interface LinearOp

Parameters:

input - the input tensor.

Returns:

the output tensor.

weight

public Tensor weight()

The weight [out, in] as an alias of the module parameter (shares storage; do not close it before the layer is done with).

Returns:

the weight, or null when the native binding is unavailable.