SAG: SQL-Retrieval Augmented Generation with Query-Time Dynamic Hyperedges
SAG retrieves structured multi-hop evidence via event-entity hyperedges and query-time joins, without a maintained global knowledge graph.
Knowledge catalogue
SAG retrieves structured multi-hop evidence via event-entity hyperedges and query-time joins, without a maintained global knowledge graph.
Karpathy’s pattern: LLMs compile raw sources into a persistent interlinked markdown wiki—entity/concept pages, index, log—rather than re-deriving answers from RAG on every query.
AD2-Bench diagnoses multimodal reasoning via a Chain of Evidence, showing failures from spatial ambiguity and semantic uncertainty in adverse urban scenes.