Enum WeightInitStrategy

pub enum WeightInitStrategy {
    KeepPretrained,
    PretrainedWithNoise {
        noise_scale: f64,
    },
    PartialInit {
        layers_to_init: Vec<String>,
    },
    Xavier,
    He,
    Custom,
}

Weight initialization strategies for transfer learning

Variants

KeepPretrained

Keep pretrained weights as-is

PretrainedWithNoise

Initialize with pretrained weights plus small noise

Fields

noise_scale: f64

Scale of noise to add to pretrained weights

PartialInit

Partial initialization (only specific layers)

Fields

layers_to_init: Vec<String>

List of layer names to initialize

Xavier

Xavier/Glorot initialization for new layers

He

He initialization for new layers

Custom

Custom initialization with user-provided function

Trait Implementations

impl Clone for WeightInitStrategy

fn clone(&self) -> WeightInitStrategy

Returns a duplicate of the value.

const fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for WeightInitStrategy

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for WeightInitStrategy

fn eq(&self, other: &WeightInitStrategy) -> bool

Tests for self and other values to be equal, and is used by ==.

const fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for WeightInitStrategy