Rethinking the Necessity of Adaptive Retrieval-Augmented Generation through the Lens of Adaptive Listwise Ranking
Jun Feng, Jiahui Tang, Zhicheng He, Hang Lv, Hongchao Gu + 3 more
TLDR
AdaRankLLM rethinks adaptive RAG, proposing a framework that optimizes retrieval for both weak and strong LLMs, significantly reducing context overhead.
Key contributions
- Proposes AdaRankLLM, an adaptive retrieval framework that re-evaluates the necessity of dynamic retrieval.
- Develops an adaptive ranker using zero-shot prompting and passage dropout for precise listwise reranking.
- Introduces a two-stage distillation paradigm to equip smaller LLMs with adaptive ranking capabilities.
- Reveals adaptive retrieval acts as a noise filter for weak models and an efficiency optimizer for strong models.
Why it matters
This paper challenges the conventional view of adaptive RAG, showing its evolving utility with LLM advancements. It provides a practical framework, AdaRankLLM, that significantly reduces context overhead while optimizing performance across diverse models. Its key insight redefines adaptive retrieval's role.
Original Abstract
Adaptive Retrieval-Augmented Generation aims to mitigate the interference of extraneous noise by dynamically determining the necessity of retrieving supplementary passages. However, as Large Language Models evolve with increasing robustness to noise, the necessity of adaptive retrieval warrants re-evaluation. In this paper, we rethink this necessity and propose AdaRankLLM, a novel adaptive retrieval framework. To effectively verify the necessity of adaptive listwise reranking, we first develop an adaptive ranker employing a zero-shot prompt with a passage dropout mechanism, and compare its generation outcomes against static fixed-depth retrieval strategies. Furthermore, to endow smaller open-source LLMs with this precise listwise ranking and adaptive filtering capability, we introduce a two-stage progressive distillation paradigm enhanced by data sampling and augmentation techniques. Extensive experiments across three datasets and eight LLMs demonstrate that AdaRankLLM consistently achieves optimal performance in most scenarios with significantly reduced context overhead. Crucially, our analysis reveals a role shift in adaptive retrieval: it functions as a critical noise filter for weaker models to overcome their limitations, while serving as a cost-effective efficiency optimizer for stronger reasoning models.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.