Struct TrainedParams 

pub struct TrainedParams {
    pub mins: Vec<f32>,
    pub maxs: Vec<f32>,
    pub dimensions: usize,
}

Trained quantization parameters computed from data

Stores per-dimension min/max values learned from a representative sample. This enables consistent quantization across all vectors and correct ADC distances.

Fields

mins: Vec<f32>

Minimum value per dimension

maxs: Vec<f32>

Maximum value per dimension

dimensions: usize

Number of dimensions

Implementations

impl TrainedParams

pub fn train(vectors: &[&[f32]]) -> Result<Self, &'static str>

Train quantization parameters from sample vectors

Computes per-dimension min/max using percentiles to exclude outliers. Uses 1st and 99th percentiles by default for robustness.

Arguments

• vectors - Sample vectors to train from (should be representative)

Errors

Returns error if vectors is empty or vectors have inconsistent dimensions.

pub fn train_with_percentiles( vectors: &[&[f32]], lower_percentile: f32, upper_percentile: f32, ) -> Result<Self, &'static str>

Train with custom percentile bounds

Arguments

• vectors - Sample vectors to train from
• lower_percentile - Lower bound percentile (e.g., 0.01 for 1st percentile)
• upper_percentile - Upper bound percentile (e.g., 0.99 for 99th percentile)

Errors

Returns error if vectors is empty or vectors have inconsistent dimensions.

pub fn quantize_value(&self, value: f32, dim: usize, levels: usize) -> u8

Quantize a single value using trained parameters for given dimension

pub fn dequantize_value(&self, code: u8, dim: usize, levels: usize) -> f32

Dequantize a code to reconstructed value for given dimension

Trait Implementations

impl Clone for TrainedParams

fn clone(&self) -> TrainedParams

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TrainedParams

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for TrainedParams

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for TrainedParams

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

where __S: Serializer,

Serialize this value into the given Serde serializer.