Vanilla RAG
The most basic retrieval-augmented generation architecture, which feeds the query and retrieved passages directly into an unconstrained large model for generation. It is simple to implement but suffers from token bloat.
Timeline (last 90 days)
Token efficiency η is defined as the ratio of effective information tokens to total output tokens, and Vanilla RAG typically has an η value below 0.5 on similar tasks.
Vanilla RAG handles simple queries efficiently but performs poorly on relational reasoning or multi-hop reasoning.
All Facts (2)
Token efficiency η is defined as the ratio of effective information tokens to total output tokens, and Vanilla RAG typically has an η value below 0.5 on similar tasks.
50%UnverifiedVanilla RAG handles simple queries efficiently but performs poorly on relational reasoning or multi-hop reasoning.
50%