Enum ScaleUpdatePolicy 

pub enum ScaleUpdatePolicy {
    Static,
    Dynamic,
    Gradual,
}

Determines how the loss scale is adjusted over training.

Variants

Static

Fixed scale – never adjusted.

Dynamic

Classic dynamic loss scaling: double after scale_up_interval clean steps, halve immediately on overflow.

Gradual

Gradual policy: multiply by scale_up_factor after scale_up_interval clean steps, multiply by scale_down_factor on overflow (both factors should be in (0, ∞), down_factor < 1).

Trait Implementations

impl Clone for ScaleUpdatePolicy

fn clone(&self) -> ScaleUpdatePolicy

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ScaleUpdatePolicy

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

Formats the value using the given formatter.

impl Display for ScaleUpdatePolicy

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

Formats the value using the given formatter.

impl Eq for ScaleUpdatePolicy

impl Hash for ScaleUpdatePolicy

fn hash<__H: Hasher>(&self, state: &mut __H)

Feeds this value into the given Hasher.

fn hash_slice<H>(data: &[Self], state: &mut H)

where H: Hasher, Self: Sized,

Feeds a slice of this type into the given Hasher.

impl PartialEq for ScaleUpdatePolicy

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

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

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 ScaleUpdatePolicy