DynaKRAG Adds Learnable Evidence Control to Multi-Hop RAG

arXiv·medium signal

DynaKRAG introduces learnable evidence control for multi-hop retrieval-augmented generation, letting the system decide which sequentially-retrieved documents to retain as new bridge facts emerge, instead of naively accumulating context. This targets a concrete pain point for RAG builders: context bloat and irrelevant-evidence drift across hops. The learnable retain/drop gate is the transferable idea.