RetroLM: Retrieval-Augmented KVs for Long-Context Processing
Kun Luo, Zheng Liu, Shitao Xiao, Jiabei Chen, Hongjin Qian, Peitian Zhang, Shanshan Jiang, Bin Dong
Abstract
Long-context processing remains a significant challenge for large language models (LLMs). Retrieval-augmented generation (RAG) has recently emerged as a promising approach, enabling LLMs to selectively access relevant information from extended contexts to improve efficiency. However, existing RAG approaches often lag behind other efficient long-context processing methods primarily due to inherent limitations on inaccurate retrieval and fragmented contexts. To address these limitations, we propose RetroLM, a novel RAG framework designed for effective long-context processing. Unlike traditional approaches, RetroLM introduces KV-level retrieval augmentation, which partitions the LLM
BibTeX
@inproceedings{aaai2026_retrolmretrieval,
title = {RetroLM: Retrieval-Augmented KVs for Long-Context Processing},
author = {Kun Luo and Zheng Liu and Shitao Xiao and Jiabei Chen and Hongjin Qian and Peitian Zhang and Shanshan Jiang and Bin Dong and Jun Zhao and Kang Liu},
booktitle = {AAAI 2026},
year = {2026}
}