![[2602.22224] DS SERVE: A Framework for Efficient and Scalable Neural Retrieval](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.22224] DS SERVE: A Framework for Efficient and Scalable Neural Retrieval
Summary
DS SERVE is a framework designed to enhance neural retrieval systems by efficiently processing large-scale text datasets, achieving low latency and supporting various inference-time trade-offs.
Why It Matters
As the demand for efficient information retrieval grows, DS SERVE addresses the challenges of latency and accuracy in neural retrieval systems. Its ability to handle vast datasets makes it relevant for applications in AI, particularly in retrieval-augmented generation and search agents, which are crucial for improving user experience in various tech applications.
Key Takeaways
- DS SERVE transforms large text datasets into high-performance retrieval systems.
- The framework supports low latency and modest memory usage on a single node.
- Inference-time trade-offs allow for customization between latency, accuracy, and result diversity.
- It is applicable for various uses, including retrieval-augmented generation and training search agents.
- The framework is expected to facilitate advancements in AI applications.
Computer Science > Information Retrieval arXiv:2602.22224 (cs) [Submitted on 17 Dec 2025] Title:DS SERVE: A Framework for Efficient and Scalable Neural Retrieval Authors:Jinjian Liu, Yichuan Wang, Xinxi Lyu, Rulin Shao, Joseph E. Gonzalez, Matei Zaharia, Sewon Min View a PDF of the paper titled DS SERVE: A Framework for Efficient and Scalable Neural Retrieval, by Jinjian Liu and 6 other authors View PDF HTML (experimental) Abstract:We present DS-Serve, a framework that transforms large-scale text datasets, comprising half a trillion tokens, into a high-performance neural retrieval system. DS-Serve offers both a web interface and API endpoints, achieving low latency with modest memory overhead on a single node. The framework also supports inference-time trade-offs between latency, accuracy, and result diversity. We anticipate that DS-Serve will be broadly useful for a range of applications, including large-scale retrieval-augmented generation (RAG), training data attribution, training search agents, and beyond. Subjects: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) Cite as: arXiv:2602.22224 [cs.IR] (or arXiv:2602.22224v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2602.22224 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Yichuan Wang [view email] [v1] Wed, 17 Dec 2025 00:43:10 UTC (856 KB) Full-text links: Access Paper: View a PDF of the paper titled DS SERVE: A Framework for Ef...