Module quantization 

Vector quantization for memory-efficient storage

This module provides various quantization methods to compress high-dimensional vectors while preserving similarity search accuracy.

Re-exports

pub use product::OptimizedProductQuantizer;

pub use product::PQCode;

pub use product::ProductQuantizer;

pub use product::ProductQuantizerConfig;

pub use product::QuantizationBenchmark;

pub use product::QuantizationBenchmarker;

pub use product::QuantizationComparison;

pub use scalar::QuantizedVector;

pub use scalar::ScalarQuantizer;

pub use scalar::ScalarQuantizerConfig;

pub use stores::BinaryVectorStore;

pub use stores::QuantizedVectorStore;

Modules

product

Product Quantization (PQ, OPQ) and benchmarking utilities.

scalar

Scalar (INT8/UINT8) vector quantization.

stores

Memory-efficient vector stores: INT8 (QuantizedVectorStore) and binary (BinaryVectorStore).

Functions

dequantize_i8_to_f32

Dequantize INT8 vector back to f32.

quantize_f32_to_i8

Quantize an f32 slice to INT8 using symmetric min-max scaling.