Struct Quantizer 

pub struct Quantizer { /* private fields */ }

Quantizer that converts f32 vectors to int8 with per-vector calibration.

Implementations

impl Quantizer

pub fn new(dimensions: usize) -> Self

Create a new quantizer for vectors of the given dimensionality.

pub fn dimensions(&self) -> usize

The configured dimensionality.

pub fn quantize(&self, vector: &[f32]) -> Result<QuantizedVector, MemoryError>

Quantize a single f32 vector to int8 with per-vector symmetric calibration.

Symmetric quantization maps to [-127, 127] (254 discrete levels). Each vector gets its own scale/zero_point derived from its min/max values.

pub fn dequantize(&self, qv: &QuantizedVector) -> Vec<f32>

Dequantize back to f32 (approximate reconstruction).

Trait Implementations

impl Clone for Quantizer

fn clone(&self) -> Quantizer

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Quantizer

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

Formats the value using the given formatter.