Trait burn_core::tensor::quantization::Quantization

pub trait Quantization<E, Q>
where
    E: Float,
    Q: PrimInt,
{
    // Required methods
    fn new(alpha: E, beta: E) -> Self;
    fn quantize(&self, values: &[E]) -> Vec<Q>;
    fn dequantize(&self, values: &[Q]) -> Vec<E>;
}

Quantization scheme to convert elements of a higher precision data type E to a lower precision data type Q and vice-versa.

Required Methods

fn new(alpha: E, beta: E) -> Self

Create a new quantization scheme for an input range [alpha, beta].

fn quantize(&self, values: &[E]) -> Vec<Q>

Convert the values to a lower precision data type.

fn dequantize(&self, values: &[Q]) -> Vec<E>

Convert the values back to a higher precision data type.

Object Safety

This trait is not object safe.

Implementors

impl<E, Q> Quantization<E, Q> for SymmetricQuantization<E, Q>

where E: Float, Q: PrimInt,

impl<E, Q, A> Quantization<E, Q> for AffineQuantization<E, Q, A>

where E: Float, Q: PrimInt, A: PrimInt,