Struct LogQuantizedVector 

pub struct LogQuantizedVector {
    pub codes: Vec<u8>,
    pub signs: Vec<bool>,
    pub max_log: f32,
    pub min_magnitude: f32,
}

Log-scaled int8 quantized vector (in-tree; no ruvector-core).

Fields

codes: Vec<u8>

Per-dimension codes 0..=255.

signs: Vec<bool>

Sign bits packed as true = negative original value.

max_log: f32

Shared scale for reconstruction.

min_magnitude: f32

Config min magnitude used at encode.

Implementations

impl LogQuantizedVector

pub fn quantize(vector: &[f32], cfg: &LogQuantizedConfig) -> Self

Quantize vector with log-scaled magnitude bins.

pub fn reconstruct(&self) -> Vec<f32>

Reconstruct approximate f32 vector.

pub fn mse(&self, original: &[f32]) -> f32

Mean squared reconstruction error vs original.

Trait Implementations

impl Clone for LogQuantizedVector

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LogQuantizedVector

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

Formats the value using the given formatter.

impl PartialEq for LogQuantizedVector

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

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for LogQuantizedVector