Adaptive-RAG is a strategy for dynamically selecting the most suitable route for a given user query. A route could be vector database retrieval, web search, direct answer, or any other viable route.

Corrective-RAG (CRAG) is a strategy for RAG that incorporates self-reflection / self-grading on retrieved documents.

So if we combine the two, we get a system that can dynamically select the most suitable route for a given user query and then self-reflect on the retrieved documents to improve the quality of the response. This is a powerful combination that can lead to more accurate and relevant responses to user queries.

We will use LanceDB as our vector database and Exa as our web search engine.

The possible routes that are guaranteed using Instructor:

1. names of all available LanceDB tables.
2. web_search
3. assistant
4. terminate (if the query is in ["exit", "quit", "q"])

So the transitions check for routes 2-4 and we assume that it must be one of the tables otherwise. ask_assistant will combine all lancedb_results and exa_search_results for context if they exist in the state, and then reset them to [] We stop after ask_assistant to get the next query for router.

Adaptive-CRAG 🧪

We have some of the core concepts from the Burr Docs in a folder and we will add that to a LanceDB table for Hybrid Search. You can add as many tables as you want. Each table will be a different route for the router.