embeddings · dim 3 · cosineSEARCH … USING VECTOR() returns internal surrogate
ids + distances (it does not project document fields), so the catalog
below is the fixed seed set. Vector search needs the pgwire
transport — over native it is gap #45 (activerecord-nodedb-adapter
docs/bugs/018).
Added vectors are searchable but not listed in the seed catalog.
| # | Title | Vector |
|---|---|---|
| 0 | Intro to AI | [0.1, 0.2, 0.3] |
| 1 | Deep Learning | [0.4, 0.5, 0.6] |
| 2 | Graph Databases | [0.9, 0.1, 0.1] |
| 3 | Vector Search | [0.15, 0.25, 0.35] |