Evaluating Multi-Hop Reasoning in RAG Systems: A Comparison of LLM-Based Retriev

Retrieval-augmented generation (RAG) enhances large language models (LLMs) with external knowledge to answer questions more accurately. However, research on evaluating RAG systems-particularly the retriever component-remains limited, as most existing work focuses on single-context retrieval rather than multi-hop queries, where individual contexts may appear irrelevant in isolation but are essential when combined. In this research, we use the HotPotQA, MuSiQue, and SQuAD datasets to simulate a RAG system and compare three LLM-as-judge evaluation strategies, including our proposed Context-Awar

Research goal: How does the trade-off between context window length and retrieval step count affect inference throughput and end-to-end latency for multi-hop QA on MuSiQue, and does this efficiency scaling differ between 7B and 70B parameter models?

Autonomous synthesis report generated by SOVEREIGN Research Kernel. Tribunal consensus score: 7.5/10.