Package org.jetbrains.kotlinx.dl.api.core.initializer

Types

Constant
Link copied to clipboard
class Constant(constantValue: Float) : Initializer

Initializer that generates tensors with constant values.

Distribution
Link copied to clipboard
enum Distribution : Enum<Distribution>

Distribution.

GlorotNormal
Link copied to clipboard
class GlorotNormal(seed: Long) : VarianceScaling

The Glorot normal initializer, also called Xavier normal initializer.

GlorotUniform
Link copied to clipboard
class GlorotUniform(seed: Long) : VarianceScaling

The Glorot uniform initializer, also called Xavier uniform initializer.

HeNormal
Link copied to clipboard
class HeNormal(seed: Long) : VarianceScaling

He normal initializer.

HeUniform
Link copied to clipboard
class HeUniform(seed: Long) : VarianceScaling

He uniform variance scaling initializer.

Identity
Link copied to clipboard
class Identity(gain: Float) : Initializer

Initializer that generates the identity matrix. This initializer is only usable for generating 2D matrices. Although identity matrices are undefined for non-square matrices an "identity" matrix is generated. E.g. 2 x 3 "identity" matrix ==> [ 1., 0., 0., 0., 1., 0.]

Initializer
Link copied to clipboard
abstract class Initializer

Initializer base class: all initializers inherit this class.

LeCunNormal
Link copied to clipboard
class LeCunNormal(seed: Long) : VarianceScaling

LeCun normal initializer.

LeCunUniform
Link copied to clipboard
class LeCunUniform(seed: Long) : VarianceScaling

LeCun uniform initializer.

Mode
Link copied to clipboard
enum Mode : Enum<Mode>

Mode.

Ones
Link copied to clipboard
class Ones : Initializer

Initializer that generates tensors initialized to 1.

Orthogonal
Link copied to clipboard
class Orthogonal(gain: Float, seed: Long) : Initializer

Initializer that generates an orthogonal matrix.

ParametrizedTruncatedNormal
Link copied to clipboard
class ParametrizedTruncatedNormal(mean: Float, stdev: Float, p1: Float, p2: Float, seed: Long) : Initializer

Initializer that generates a parametrized truncated normal distribution.

RandomNormal
Link copied to clipboard
class RandomNormal(mean: Float, stdev: Float, seed: Long) : Initializer

Initializer that generates tensors with a normal distribution.

RandomUniform
Link copied to clipboard
class RandomUniform(maxVal: Float, minVal: Float, seed: Long) : Initializer

Initializer that generates tensors with a uniform distribution.

TruncatedNormal
Link copied to clipboard
class TruncatedNormal(seed: Long) : Initializer

Initializer that generates a truncated normal distribution.

VarianceScaling
Link copied to clipboard
open class VarianceScaling(scale: Double, mode: Mode, distribution: Distribution, seed: Long) : Initializer

Initializer capable of adapting its scale to the shape of weights tensors.

Zeros
Link copied to clipboard
class Zeros : Initializer

Initializer that generates tensors initialized to 0.