Package ai.djl.nn.convolutional

Class Convolution.ConvolutionBuilder<T extends Convolution.ConvolutionBuilder>

java.lang.Object

ai.djl.nn.convolutional.Convolution.ConvolutionBuilder<T>

Type Parameters:

T - the type of Convolution block to build

Direct Known Subclasses:

Conv1d.Builder, Conv2d.Builder, Conv3d.Builder

Enclosing class:

Convolution

public abstract static class Convolution.ConvolutionBuilder<T extends Convolution.ConvolutionBuilder>
extends java.lang.Object

A builder that can build any Convolution block.

Field Summary

Fields

Modifier and Type	Field	Description
protected Shape	dilation
protected int	filters
protected int	groups
protected boolean	includeBias
protected Shape	kernelShape
protected Shape	padding
protected Shape	stride

Constructor Summary

Constructors

Constructor	Description
ConvolutionBuilder()

Method Summary

Modifier and Type	Method	Description
T	optBias(boolean includeBias)	Sets the optional parameter of whether to include a bias vector.
T	optDilation(Shape dilate)	Sets the dilation along each dimension.
T	optGroups(int groups)	Sets the number of group partitions.
T	optPadding(Shape padding)	Sets the padding along each dimension.
T	optStride(Shape stride)	Sets the stride of the convolution.
protected abstract T	self()
T	setFilters(int filters)	Sets the Required number of filters.
T	setKernelShape(Shape kernelShape)	Sets the shape of the kernel.
protected void	validate()	Validates that the required arguments are set.

Methods inherited from class java.lang.Object

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

Field Detail

kernelShape

protected Shape kernelShape

stride

protected Shape stride

padding

protected Shape padding

dilation

protected Shape dilation

filters

protected int filters

groups

protected int groups

includeBias

protected boolean includeBias

Constructor Detail

ConvolutionBuilder

public ConvolutionBuilder()

Method Detail

setKernelShape

public T setKernelShape(Shape kernelShape)

Sets the shape of the kernel.

Parameters:

kernelShape - the shape of the kernel

Returns:

this Builder

optStride

public T optStride(Shape stride)

Sets the stride of the convolution. Defaults to 1 in each dimension.

Parameters:

stride - the shape of the stride

Returns:

this Builder

optPadding

public T optPadding(Shape padding)

Sets the padding along each dimension. Defaults to 0 along each dimension.

Parameters:

padding - the shape of padding along each dimension

Returns:

this Builder

optDilation

public T optDilation(Shape dilate)

Sets the dilation along each dimension. Defaults to 1 along each dimension.

Parameters:

dilate - the shape of dilation along each dimension

Returns:

this Builder

setFilters

public T setFilters(int filters)

Sets the Required number of filters.

Parameters:

filters - the number of convolution filters(channels)

Returns:

this Builder

optGroups

public T optGroups(int groups)

Sets the number of group partitions.

Parameters:

groups - the number of group partitions

Returns:

this Builder

optBias

public T optBias(boolean includeBias)

Sets the optional parameter of whether to include a bias vector. Includes bias by default.

Parameters:

includeBias - whether to use a bias vector parameter

Returns:

this Builder

validate

protected void validate()

Validates that the required arguments are set.

Throws:

java.lang.IllegalArgumentException - if the required arguments are not set

self

protected abstract T self()