Struct QuantizedWeights 

pub struct QuantizedWeights {
    pub format: QuantFormat,
    pub data: Vec<u8>,
    pub shape: (usize, usize),
    pub layer_name: String,
}

Quantized weight storage for a single layer.

Fields

format: QuantFormat

Quantization format

data: Vec<u8>

Raw quantized data

shape: (usize, usize)

Shape: [rows, cols] for 2D weights

layer_name: String

Layer name (for debugging)

Implementations

impl QuantizedWeights

pub fn new( format: QuantFormat, data: Vec<u8>, shape: (usize, usize), name: &str, ) -> Self

Create new quantized weights.

pub fn num_weights(&self) -> usize

Total number of weights.

pub fn memory_bytes(&self) -> usize

Memory footprint in bytes.

pub fn f16_memory_bytes(&self) -> usize

Memory footprint if stored as F16.

pub fn compression_ratio(&self) -> f64

Compression ratio (F16 / quantized).

pub fn actual_bits_per_weight(&self) -> f64

Effective bits per weight (actual).

Trait Implementations

impl Clone for QuantizedWeights

fn clone(&self) -> QuantizedWeights

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for QuantizedWeights

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

Formats the value using the given formatter.