Struct IvfConfig 

pub struct IvfConfig {
    pub n_list: usize,
    pub nprobe: usize,
    pub training_sample_size: usize,
    pub kmeans_iters: usize,
    pub seed: Option<u64>,
}

Configuration for an IVF (Inverted File) index.

IVF partitions the vector space into n_list Voronoi cells via k-means. At query time, only the nprobe cells closest to the query are searched, making search cost roughly O(n_list + nprobe * N / n_list) instead of O(N). Higher nprobe yields higher recall at the cost of latency.

Defaults follow the FAISS convention of n_list ≈ sqrt(N) and nprobe = n_list / 16 for a ~0.95 recall@10 starting point.

Fields

n_list: usize

Number of Voronoi cells (centroids).

nprobe: usize

Number of cells to probe per query.

training_sample_size: usize

Upper bound on training-sample size used by k-means. The effective sample is min(training_sample_size, input.len()).

kmeans_iters: usize

Maximum Lloyd iterations during training.

seed: Option<u64>

Random seed for k-means++ initialisation. None ⇒ non-deterministic.

Implementations

impl IvfConfig

pub fn for_dataset_size(n: usize) -> Self

Heuristic n_list ≈ sqrt(N) clamped to a reasonable range.

Trait Implementations

impl Clone for IvfConfig

fn clone(&self) -> IvfConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for IvfConfig

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

Formats the value using the given formatter.

impl Default for IvfConfig

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for IvfConfig

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for IvfConfig

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

where __S: Serializer,

Serialize this value into the given Serde serializer.