Struct VectorConfig 

pub struct VectorConfig {
    pub dims: u32,
    pub index_kind: IndexKind,
    pub metric: MetricKind,
    pub quantization: ScalarKind,
    pub connectivity: usize,
    pub expansion_add: usize,
    pub expansion_search: usize,
    pub ivf_nlist: usize,
    pub ivf_nprobe: usize,
}

Tunable parameters for the vector index. Built from CLI flags; VectorConfig::new supplies the defaults used by tests and any caller that only cares about the embedding dimension.

Fields

dims: u32

Embedding dimension. All upserted/queried vectors must match this.

index_kind: IndexKind

Which ANN backend to use.

metric: MetricKind

Distance metric used by the index.

quantization: ScalarKind

On-disk/in-index scalar representation (enables quantization). HNSW only.

connectivity: usize

HNSW graph degree (M). Higher = better recall, more memory.

expansion_add: usize

HNSW efConstruction. Higher = better index quality, slower inserts.

expansion_search: usize

HNSW efSearch. Higher = better recall, slower queries.

ivf_nlist: usize

IVF number of Voronoi cells (centroids). IVF only.

ivf_nprobe: usize

IVF default cells probed per query. IVF only.

Implementations

impl VectorConfig

pub const fn new(dims: u32) -> Self

Default HNSW configuration for the given embedding dimension.

Trait Implementations

impl Clone for VectorConfig

fn clone(&self) -> VectorConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for VectorConfig

impl Debug for VectorConfig

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

Formats the value using the given formatter.