Threadbare implementation of a RAG toolkit:
- document chunking
- fixed length
- Recursive Character Split (RCS)
- document specific splitting
- semantic splitting
- document retrieval (exhaustive search for a one-liner "vector database")
- Approximate Nearest Neighbor (ANN) with Navigable Small World (NSW) for K-Nearest Neighbor (KNN).