Module l2 

L2 semantic cache. L2 semantic cache (embedding + vector search + rescoring).

The default implementation embeds with SinterEmbedder, searches a binary-quantized backend, then rescoring top candidates in full precision.

Re-exports

pub use backend::BqSearchBackend;

pub use cache::L2SemanticCache;

pub use cache::L2SemanticCacheHandle;

pub use config::DEFAULT_TOP_K_BQ;

pub use config::DEFAULT_TOP_K_FINAL;

pub use config::L2_COLLECTION_NAME;

pub use config::L2_VECTOR_SIZE;

pub use config::L2Config;

pub use error::L2CacheError;

pub use error::L2CacheResult;

pub use loader::MockStorageLoader;mock

pub use loader::NvmeStorageLoader;

pub use loader::StorageLoader;

pub use mock::MockL2SemanticCache;mock

pub use types::L2LookupResult;

Modules

backend

Backend trait used by L2 for vector search/upsert.

cache

Core L2 cache implementation.

config

L2 configuration.

error

L2 error types.

loader

Storage loader traits and implementations.

mockmock

Mock L2 cache helpers (enabled with mock feature). Mock L2 cache helpers (in-memory storage + mock vector DB).

types

L2 result types.