Struct QuantizedTensor

pub struct QuantizedTensor {
    pub data: ArrayD<i8>,
    pub params: QuantizationParams,
    pub shape: Vec<usize>,
}

Quantized tensor representation

Fields

data: ArrayD<i8>

Quantized integer data

params: QuantizationParams

Quantization parameters

shape: Vec<usize>

Original tensor shape

Implementations

impl QuantizedTensor

pub fn from_float( tensor: &ArrayD<f32>, config: &QuantizationConfig, ) -> Result<Self>

Create new quantized tensor from float tensor

pub fn dequantize(&self) -> ArrayD<f32>

Dequantize back to float tensor

pub fn size_bytes(&self) -> usize

Get quantized tensor size in bytes

pub fn compression_ratio(&self) -> f32

Get compression ratio compared to f32

Trait Implementations

impl Clone for QuantizedTensor

fn clone(&self) -> QuantizedTensor

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for QuantizedTensor

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

Formats the value using the given formatter.