Repository navigation
#417 Embedding LRU cache for repeated queries #420
Description
Problem
Query embeddings recomputed every time (cheap with hash trick, but wasteful).
Location
src/index/brain.rs:190
Solution
Add LRU cache (64 entries) for query to embedding.
Expected Impact
Marginal. Future-proofing for real model embeddings.
Activity
ajianaz commentedon Jul 28, 2026
Meta-tracker. All phase-3 sub-issues (#417, #416, #415, #414, #413) either closed or deferred. Low priority, closing tracker.