Struct LoadedIndex 

pub struct LoadedIndex {
    pub metadata: Metadata,
    pub codec: ResidualCodec,
    pub ivf: IvfStridedTensor,
    pub doc_codes: StridedTensor<usize>,
    pub doc_residuals: StridedTensor<u8>,
    pub nbits: usize,
}

A loaded index optimized for search with StridedTensor storage.

This struct contains all data required for search operations, stored in a format optimized for batch lookups. It uses StridedTensor for efficient retrieval of variable-length sequences.

Fields

metadata: Metadata

Index metadata

codec: ResidualCodec

Residual codec for quantization/decompression

ivf: IvfStridedTensor

IVF index mapping centroids to document IDs

doc_codes: StridedTensor<usize>

Document codes (centroid assignments) stored contiguously

doc_residuals: StridedTensor<u8>

Packed residuals stored contiguously

nbits: usize

Number of bits for quantization

Implementations

impl LoadedIndex

pub fn from_index(index: Index) -> Self

Create a LoadedIndex from an existing Index.

This converts the Index’s per-document storage to contiguous StridedTensor storage.

pub fn get_candidates(&self, centroid_indices: &[usize]) -> Vec<i64>

Get candidate documents from IVF for given centroid indices.

pub fn lookup_documents( &self, doc_ids: &[usize], ) -> (Array1<usize>, Array2<u8>, Array1<i64>)

Lookup codes and residuals for a batch of document IDs.

Returns (codes, residuals, lengths) for efficient batch decompression.

pub fn decompress_documents( &self, doc_ids: &[usize], ) -> Result<(Array2<f32>, Array1<i64>)>

Decompress embeddings for a batch of document IDs.

Returns the decompressed embeddings as a contiguous array along with lengths.

pub fn num_documents(&self) -> usize

Get the number of documents in the index.

pub fn embedding_dim(&self) -> usize

Get the embedding dimension.

pub fn reconstruct(&self, doc_ids: &[i64]) -> Result<Vec<Array2<f32>>>

Reconstruct embeddings for specific documents.

This method retrieves the compressed codes and residuals for each document and decompresses them to recover the original embeddings.

Arguments

• doc_ids - Slice of document IDs to reconstruct (0-indexed)

Returns

A vector of 2D arrays, one per document. Each array has shape [num_tokens, dim].

Example

use next_plaid::LoadedIndex;

let index = LoadedIndex::load("/path/to/index")?;
let embeddings = index.reconstruct(&[0, 1, 2])?;

for (i, emb) in embeddings.iter().enumerate() {
    println!("Document {}: {} tokens x {} dim", i, emb.nrows(), emb.ncols());
}

pub fn reconstruct_single(&self, doc_id: i64) -> Result<Array2<f32>>

Reconstruct a single document’s embeddings.

Convenience method for reconstructing a single document.

Arguments

• doc_id - Document ID to reconstruct (0-indexed)

Returns

A 2D array with shape [num_tokens, dim].