Documents split into chunks with vector embeddings, HNSW similarity search and trigram text search.
6 tables · 6 relationships