core/org.gnit.lucenekmp.codecs.lucene102

Package-level declarations

Types

BinarizedByteVectorValues

abstract class BinarizedByteVectorValues : ByteVectorValues

Binarized byte vector values

Lucene102BinaryFlatVectorsScorer

class Lucene102BinaryFlatVectorsScorer(nonQuantizedDelegate: FlatVectorsScorer) : FlatVectorsScorer

Vector scorer over binarized vector values

Lucene102BinaryQuantizedVectorsFormat

class Lucene102BinaryQuantizedVectorsFormat : FlatVectorsFormat

The binary quantization format used here is a per-vector optimized scalar quantization. These ideas are evolutions of LVQ proposed in Similarity search in the blink of an eye with compressed indices by Cecilia Aguerrebere et al., the previous work on globally optimized scalar quantization in Apache Lucene, and Accelerating Large-Scale Inference with Anisotropic Vector Quantization by Ruiqi Guo et. al. Also see [ ]. Some of key features are:

Lucene102BinaryQuantizedVectorsWriter

class Lucene102BinaryQuantizedVectorsWriter(vectorsScorer: Lucene102BinaryFlatVectorsScorer, rawVectorDelegate: FlatVectorsWriter, state: SegmentWriteState) : FlatVectorsWriter

Copied from Lucene, replace with Lucene's implementation sometime after Lucene 10

Lucene102HnswBinaryQuantizedVectorsFormat

class Lucene102HnswBinaryQuantizedVectorsFormat(maxConn: Int = Lucene99HnswVectorsFormat.DEFAULT_MAX_CONN, beamWidth: Int = Lucene99HnswVectorsFormat.DEFAULT_BEAM_WIDTH, numMergeWorkers: Int = Lucene99HnswVectorsFormat.DEFAULT_NUM_MERGE_WORKER, mergeExec: ExecutorService? = null) : KnnVectorsFormat

A vectors format that uses HNSW graph to store and search for vectors. But vectors are binary quantized using Lucene102BinaryQuantizedVectorsFormat before being stored in the graph.

OffHeapBinarizedVectorValues

abstract class OffHeapBinarizedVectorValues : BinarizedByteVectorValues

Binarized vector values loaded from off-heap