Struct QuantizedTensor 

pub struct QuantizedTensor {
    pub data: Vec<i8>,
    pub scales: Vec<f32>,
    pub zero_points: Vec<i8>,
    pub shape: Vec<usize>,
    pub scheme: QuantizationScheme,
}

Quantized tensor representation

Fields

data: Vec<i8>

Quantized values (INT8 or FP16)

scales: Vec<f32>

Scale factors for dequantization

zero_points: Vec<i8>

Zero points for asymmetric quantization

shape: Vec<usize>

Original shape

scheme: QuantizationScheme

Quantization scheme used

Implementations

impl QuantizedTensor

pub fn from_tensor(tensor: &Tensor, config: &QuantizationConfig) -> Self

Create a new quantized tensor from float tensor

pub fn dequantize(&self) -> Tensor

Dequantize back to float tensor

pub fn compression_ratio(&self) -> f32

Get compression ratio

Trait Implementations

impl Clone for QuantizedTensor

fn clone(&self) -> QuantizedTensor

Returns a duplicate of the value.

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.