Evaluation Methods

Two arXiv Preprints Propose Adaptive RAG With Complexity Scoring and Agentic Retrieval

The authors report benchmark gains pending independent validation as arXiv preprints.