Enum WeightQuantization 

#[repr(u8)]
pub enum WeightQuantization {
    Float32 = 0,
    Float16 = 1,
    UInt8 = 2,
    UInt4 = 3,
}

Quantization format for sparse vector weights

Float32 preserves input precision. Smaller weight representations trade precision for payload size; retrieval quality depends on the workload and must be measured. Integer encodings also store per-vector scale and offset, so total index-size savings differ from the ratio of weight widths.

Variants

Float32 = 0

Full 32-bit float precision

Float16 = 1

16-bit float (half precision), two bytes per weight

UInt8 = 2

8-bit integer codes with per-vector scale and offset

UInt4 = 3

4-bit integer codes (two per byte) with per-vector scale and offset

Implementations

impl WeightQuantization

pub fn bytes_per_weight(&self) -> f32

Bytes per weight (approximate for UInt4)

Trait Implementations

impl Clone for WeightQuantization

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for WeightQuantization

impl Debug for WeightQuantization

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

Formats the value using the given formatter.

impl Default for WeightQuantization

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for WeightQuantization

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Eq for WeightQuantization

impl PartialEq for WeightQuantization

fn eq(&self, other: &Self) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for WeightQuantization

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for WeightQuantization