Struct Dropout

pub struct Dropout {
    pub dropout_rate: f64,
    pub is_training: bool,
    pub variational: bool,
    /* private fields */
}

Dropout layer for regularization

Implements different types of dropout:

• Standard dropout: randomly sets elements to zero
• Variational dropout: uses same mask across time steps (for RNNs)
• Zoneout: keeps some hidden/cell state values from previous timestep

Fields

dropout_rate: f64

is_training: bool

variational: bool

Implementations

impl Dropout

pub fn new(dropout_rate: f64) -> Self

pub fn variational(dropout_rate: f64) -> Self

pub fn train(&mut self)

pub fn eval(&mut self)

pub fn forward(&mut self, input: &Array2<f64>) -> Array2<f64>

pub fn get_last_mask(&self) -> Option<&Array2<f64>>

pub fn backward(&self, grad_output: &Array2<f64>) -> Array2<f64>

Trait Implementations

impl Clone for Dropout

fn clone(&self) -> Dropout

Returns a duplicate of the value.

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

Performs copy-assignment from source.