Crate tokenmiser_cache 

Two-layer cache: L1Cache is an exact-match TTL LRU, L2Cache is a per-tenant semantic cache over bge-small embeddings.

Re-exports

pub use key::exact_key;

pub use l2::L2Cache;

pub use l2::SemanticStats;

Modules

key

L1 exact-match cache key derivation.

l2

L2 semantic cache: prompts embedded with bge-small-en-v1.5 and matched per-tenant by cosine similarity against a threshold.

Structs

CacheStats

L1Cache

Exact-match cache: a capacity-bounded LRU with per-entry TTL. A single mutex suffices because the critical section is a hash lookup plus a clone.