Struct GradientDelta 

pub struct GradientDelta {
    pub values: Vec<f32>,
    pub is_full: bool,
    pub round: u64,
}

A gradient update that is either a complete gradient or a delta from the previous round.

Fields

values: Vec<f32>

Either the full gradient vector (is_full = true) or the element-wise delta from the previous round’s gradient.

is_full: bool

true when values contains the full gradient rather than a delta.

round: u64

The federated learning round number this update belongs to.

Implementations

impl GradientDelta

pub fn compression_ratio(&self, original_len: usize) -> f64

Compression ratio expressed as mean absolute delta divided by the maximum possible delta magnitude.

For full gradients, returns 1.0 (no compression). For delta updates, a lower value indicates a smaller delta and therefore better effective compression relative to sending the full gradient.

original_len is the number of elements in the underlying gradient (used for normalisation when values.len() != original_len).

Trait Implementations

impl Clone for GradientDelta

fn clone(&self) -> GradientDelta

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for GradientDelta

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for GradientDelta

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for GradientDelta

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.