Struct EmbeddingNormalizer 

pub struct EmbeddingNormalizer { /* private fields */ }

Embedding normalizer with configurable strategies and statistics tracking.

Implementations

impl EmbeddingNormalizer

pub fn new(config: NormalizerConfig) -> Self

Create a new normalizer with the given configuration.

pub fn normalize(&mut self, embedding: &mut [f64]) -> NormStats

Normalize a single embedding vector in-place, returning statistics.

pub fn normalize_batch(&mut self, embeddings: &mut [Vec<f64>]) -> Vec<NormStats>

Normalize a batch of embeddings, returning per-vector statistics.

pub fn l1_norm(v: &[f64]) -> f64

Compute the L1 (Manhattan) norm of a vector.

pub fn l2_norm(v: &[f64]) -> f64

Compute the L2 (Euclidean) norm of a vector.

pub fn linf_norm(v: &[f64]) -> f64

Compute the L-infinity norm (maximum absolute value) of a vector.

pub fn compute_mean(v: &[f64]) -> f64

Compute the arithmetic mean of a vector.

pub fn compute_std_dev(v: &[f64], mean: f64) -> f64

Compute the population standard deviation given a precomputed mean.

pub fn clip(&self, embedding: &mut [f64])

Clip embedding values to the configured bounds (if any).

pub fn cosine_similarity(a: &[f64], b: &[f64]) -> f64

Compute cosine similarity between two vectors.

Returns 0.0 if either vector has zero norm.

pub fn compute_norm_stats(original: &[f64], normalized: &[f64]) -> NormStats

Compute comprehensive statistics comparing original and normalized vectors.

pub fn stats(&self) -> &NormalizerStats

Get a reference to the aggregate normalizer statistics.

pub fn reset_stats(&mut self)

Reset all aggregate statistics to their defaults.