Struct Int4Quantized 

pub struct Int4Quantized {
    pub data: Vec<u8>,
    pub min: f32,
    pub scale: f32,
    pub dimensions: usize,
}

Int4 quantization (8x compression)

Quantizes f32 to 4-bit integers (0-15), packing 2 values per byte. Provides 8x compression with better precision than binary.

Fields

data: Vec<u8>

Packed 4-bit values (2 per byte)

min: f32

Minimum value for dequantization

scale: f32

Scale factor for dequantization

dimensions: usize

Number of dimensions

Implementations

impl Int4Quantized

pub fn quantize(vector: &[f32]) -> Self

Quantize a vector to 4-bit representation

pub fn distance(&self, other: &Self) -> f32

Calculate distance to another Int4 quantized vector

pub fn reconstruct(&self) -> Vec<f32>

Reconstruct approximate full-precision vector

pub fn compression_ratio() -> f32

Get compression ratio (8x for Int4)

Trait Implementations

impl Clone for Int4Quantized

fn clone(&self) -> Int4Quantized

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Int4Quantized

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for Int4Quantized

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for Int4Quantized

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

where __S: Serializer,

Serialize this value into the given Serde serializer.