Join the discussion
Write your take first — we'll ask for email only when you're ready to publish.
- Hacker News
- storage utilisation is substantial. store both inverted indexes and dense embeddings creates significant overhead. with 1k dimension vectors, an average like 20 page doc with 30-100 chunks will be 120-400kb. HNSW tops with 1.5-2.0x the size of db.
although trading disk space for sub second search is good choice. overall great technical execution!
by azterizm