Meet LEANN Open-Source
The headline numbers are striking, but the real story is the trade: LEANN buys a ~30x storage cut by re-deriving embeddings at query time, so the bill simply moves from RAM to latency. Seconds-level search rules it out for interactive retrieval, yet for personal knowledge bases, edge deployments and archival RAG it could genuinely matter.… more
In our view, the HNSW accuracy-parity claim deserves independent replication before anyone re-architects a production stack — author-run benchmarks on self-selected workloads tend to flatter the new method. Still, an open-source Berkeley release in a space where vector storage is becoming a real line item is worth watching.
Meet LEANN: An Open-Source Vector Index that Shrinks a 60M-Chunk RAG Index from 201 GB to 6 GB
UC Berkeley’s LEANN vector index stores a pruned graph and recomputes embeddings at query time. It matches HNSW accuracy at 90% recall using 4 GB instead of 188 GB, at the cost of seconds-level search. The post Meet LEANN: An Open-Source Vector Index tha…
Source:
Marktechpost
Discussion (0)