Enum W8a8Scale 

pub enum W8a8Scale {
    PerRow,
    Block128,
}

2026-09-28: How a W8A8 weight carries its scales, and so how the activation is quantized: one scale per output row with one per token, or one per 128x128 weight block with one per (token, 128-wide K group).

Variants

PerRow

Block128

Implementations

impl W8a8Scale

pub fn of(format: WeightQuantFormat) -> Option<Self>

2026-09-28: The layout of format, or None for a format this family does not read.

pub fn act_scales_per_row(self, k: u32) -> usize

2026-09-28: FP32 activation scales per row for a K-wide activation.

Trait Implementations

impl Clone for W8a8Scale

fn clone(&self) -> W8a8Scale

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for W8a8Scale

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

Formats the value using the given formatter.

impl PartialEq for W8a8Scale

fn eq(&self, other: &W8a8Scale) -> 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 Copy for W8a8Scale

impl Eq for W8a8Scale

impl StructuralPartialEq for W8a8Scale