Enum WeightInit 

pub enum WeightInit {
    Zero,
    Random,
    Xavier,
    He,
    Uniform {
        low: f64,
        high: f64,
    },
    Normal {
        mean: f64,
        std: f64,
    },
}

Weight initialization strategies

Variants

Zero

Initialize all weights to zero (not recommended for most layers)

Random

Uniform random initialization in (-1, 1) without any scaling

Xavier

Xavier/Glorot initialization: scales by sqrt(2 / (fan_in + fan_out))

He

He/Kaiming initialization: scales by sqrt(2 / fan_in), suited for ReLU activations

Uniform

Uniform distribution over [low, high)

Fields

low: f64

Lower bound of the uniform distribution

high: f64

Upper bound of the uniform distribution

Normal

Gaussian distribution with specified mean and standard deviation

Fields

mean: f64

Mean of the Gaussian distribution

std: f64

Standard deviation of the Gaussian distribution

Trait Implementations

impl Clone for WeightInit

fn clone(&self) -> WeightInit

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for WeightInit

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

Formats the value using the given formatter.

impl PartialEq for WeightInit

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

Equality operator ==.

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

Inequality operator !=.

impl StructuralPartialEq for WeightInit