Struct VqCodebook 

pub struct VqCodebook { /* private fields */ }

Vector quantization codebook for bootstrapping new entity embeddings.

When a new entity type has no training data for embeddings, the codebook provides a reasonable starting vector by mapping entity types to their nearest centroid from previously seen embeddings.

Implementations

impl VqCodebook

pub fn new(k: usize, dim: usize) -> Self

Create a new codebook with k centroids of dimension dim.

Centroids are initialized to zero and must be trained before use.

pub fn k(&self) -> usize

Number of centroids in the codebook.

pub fn dim(&self) -> usize

Dimensionality of the embedding space.

pub fn train(&mut self, embeddings: &[(String, Vec<f32>)])

Train the codebook from existing embeddings using Lloyd’s algorithm.

Each embedding is a (entity_type, vector) pair. After training, each entity type is assigned to its nearest centroid.

Uses K-means++ initialization for better convergence.

pub fn lookup(&self, entity_type: &str) -> Option<&[f32]>

Look up the codebook vector for an entity type.

Returns None if the entity type has not been seen during training.

pub fn quantize(&self, vector: &[f32]) -> (usize, f32)

Quantize a vector to its nearest centroid.

Returns (centroid_index, squared_distance).

Trait Implementations

impl Clone for VqCodebook

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for VqCodebook

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for VqCodebook

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for VqCodebook

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

where __S: Serializer,

Serialize this value into the given Serde serializer.