ACL 2025long0 citations

The Efficiency vs. Accuracy Trade-off: Optimizing RAG-Enhanced LLM Recommender Systems Using Multi-Head Early Exit

Huixue Zhou, Hengrui Gu, Zaifu Zhan, Xi Liu, Kaixiong Zhou, Yongkang Xiao, Mingfu Liang, Srinivas Prasad Govindan

Abstract

The deployment of Large Language Models (LLMs) in recommender systems for Click-Through Rate (CTR) prediction requires a careful balance between computational efficiency and predictive accuracy. This paper introduces OptiRAG-Rec, a comprehensive framework that integrates Retrieval-Augmented Generation (RAG) with a novel multi-head early exit architecture to address both challenges. By leveraging Graph Convolutional Networks (GCNs) as efficient retrieval mechanisms, the framework significantly reduces data retrieval times while maintaining high model performance. Additionally, the multi-head early exit strategy dynamically terminates inference based on real-time predictive confidence assessments, enhancing responsiveness without sacrificing accuracy. Experimental results demonstrate that OptiRAG-Rec reduces computation time while preserving the precision required for reliable recommendations, establishing a new benchmark for efficient and accurate LLM deployment in recommendation.

BibTeX
@inproceedings{zhou-etal-2025-efficiency,
    title = "The Efficiency vs. Accuracy Trade-off: Optimizing {RAG}-Enhanced {LLM} Recommender Systems Using Multi-Head Early Exit",
    author = "Zhou, Huixue  and
      Gu, Hengrui  and
      Zhan, Zaifu  and
      Liu, Xi  and
      Zhou, Kaixiong  and
      Xiao, Yongkang  and
      Liang, Mingfu  and
      Govindan, Srinivas Prasad  and
      Chawla, Piyush  and
      Yang, Jiyan  and
      Meng, Xiangfei  and
      Li, Huayu  and
      Zhang, Buyun  and
      Luo, Liang  and
      Chen, Wen-Yen  and
      Han, Yiping  and
      Long, Bo  and
      Zhang, Rui  and
      Chen, Tianlong",
    editor = "Che, Wanxiang  and
      Nabende, Joyce  and
      Shutova, Ekaterina  and
      Pilehvar, Mohammad Taher",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2025.acl-long.1283/",
    doi = "10.18653/v1/2025.acl-long.1283",
    pages = "26443--26458",
    ISBN = "979-8-89176-251-0"
}
The Efficiency vs. Accuracy Trade-off: Optimizing RAG-Enhanced LLM Recommender Systems Using Multi-Head Early Exit · ACL 2025