Struct GcGradientTensor 

pub struct GcGradientTensor {
    pub layer_id: String,
    pub values: Vec<f64>,
    pub step: u64,
    pub norm: f64,
}

A gradient tensor associated with a named layer at a particular training step.

norm is always the L2 norm of values and is stored pre-computed to avoid redundant recomputation during statistics gathering.

Fields

layer_id: String

Identifier for the model layer this gradient belongs to.

values: Vec<f64>

Raw gradient values (one entry per parameter).

step: u64

The global training step at which this gradient was computed.

norm: f64

Pre-computed L2 norm of values.

Implementations

impl GcGradientTensor

pub fn new(layer_id: impl Into<String>, values: Vec<f64>, step: u64) -> Self

Construct a new GcGradientTensor, computing the L2 norm automatically.

pub fn compute_norm(values: &[f64]) -> f64

Compute the L2 (Euclidean) norm of a gradient value slice.

Returns sqrt(sum(v^2)).

Trait Implementations

impl Clone for GcGradientTensor

fn clone(&self) -> GcGradientTensor

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for GcGradientTensor

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

Formats the value using the given formatter.

impl PartialEq for GcGradientTensor

fn eq(&self, other: &GcGradientTensor) -> 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 GcGradientTensor