mnem-ann

Approximate-nearest-neighbour (HNSW) vector index for mnem. Feature-gated alternative to the built-in brute-force index in mnem-core.

Implements the VectorIndex trait from mnem-core with an HNSW backend. The built-in BruteForceVectorIndex is 100%-recall and O(n x dim) per query; HnswVectorIndex trades ~1% recall for O(log n x dim) queries and is the right choice once a repo grows past roughly 10k indexed vectors or for long-lived servers where query latency dominates. Both impls return Vec<VectorHit> sorted by descending score with NodeId-ASC tiebreak, so replay stays byte-stable either way. Kept in a separate crate because most HNSW implementations carry SIMD intrinsics or architecture-specific unsafe blocks that would fail mnem-core's #![forbid(unsafe_code)] and WASM targets.

use mnem_ann::HnswVectorIndex;
use mnem_core::index::vector::VectorIndex;

let idx = HnswVectorIndex::build_from_repo(&repo, "openai:text-embedding-3-small")?;
let hits = idx.search(&query_vec, 10)?;

Workspace top: ../../README.md.

Licensed under Apache-2.0.