@inproceedings{lu-etal-2025-turborag,
title = "{T}urbo{RAG}: Accelerating Retrieval-Augmented Generation with Precomputed {KV} Caches for Chunked Text",
author = "Lu, Songshuo and
Wang, Hua and
Rong, Yutian and
Chen, Zhi and
Tang, Yaohua",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.emnlp-main.334/",
pages = "6599--6612",
ISBN = "979-8-89176-332-6",
abstract = "Current Retrieval-Augmented Generation (RAG) systems concatenate and process numerous retrieved document chunks for prefill which requires a large volume of computation, therefore leading to significant latency in time-to-first-token (TTFT). To reduce the computation overhead as well as TTFT, we introduce TurboRAG, a hybrid offline{--}online paradigm that (i) pre{-}computes chunk{-}level key-value (KV) caches, (ii) stitches them together at inference time using independent{--}attention and reordered{-}RoPE techniques, and (iii) preserves answer quality without changing the model architecture. Hence, online computation of KV caches is eliminated during inference. Our approach is applicable to most existing large language models and their applications without any requirement in modification of models and inference systems. Experimental results across a suite of RAG benchmarks demonstrate that TurboRAG reduces TTFT by up to 9.4x compared to the conventional RAG systems (on an average of 8.6x), but reserving comparable performance to the standard RAG systems."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lu-etal-2025-turborag">
<titleInfo>
<title>TurboRAG: Accelerating Retrieval-Augmented Generation with Precomputed KV Caches for Chunked Text</title>
</titleInfo>
<name type="personal">
<namePart type="given">Songshuo</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hua</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yutian</namePart>
<namePart type="family">Rong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhi</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yaohua</namePart>
<namePart type="family">Tang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-332-6</identifier>
</relatedItem>
<abstract>Current Retrieval-Augmented Generation (RAG) systems concatenate and process numerous retrieved document chunks for prefill which requires a large volume of computation, therefore leading to significant latency in time-to-first-token (TTFT). To reduce the computation overhead as well as TTFT, we introduce TurboRAG, a hybrid offline–online paradigm that (i) pre-computes chunk-level key-value (KV) caches, (ii) stitches them together at inference time using independent–attention and reordered-RoPE techniques, and (iii) preserves answer quality without changing the model architecture. Hence, online computation of KV caches is eliminated during inference. Our approach is applicable to most existing large language models and their applications without any requirement in modification of models and inference systems. Experimental results across a suite of RAG benchmarks demonstrate that TurboRAG reduces TTFT by up to 9.4x compared to the conventional RAG systems (on an average of 8.6x), but reserving comparable performance to the standard RAG systems.</abstract>
<identifier type="citekey">lu-etal-2025-turborag</identifier>
<location>
<url>https://aclanthology.org/2025.emnlp-main.334/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>6599</start>
<end>6612</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TurboRAG: Accelerating Retrieval-Augmented Generation with Precomputed KV Caches for Chunked Text
%A Lu, Songshuo
%A Wang, Hua
%A Rong, Yutian
%A Chen, Zhi
%A Tang, Yaohua
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-332-6
%F lu-etal-2025-turborag
%X Current Retrieval-Augmented Generation (RAG) systems concatenate and process numerous retrieved document chunks for prefill which requires a large volume of computation, therefore leading to significant latency in time-to-first-token (TTFT). To reduce the computation overhead as well as TTFT, we introduce TurboRAG, a hybrid offline–online paradigm that (i) pre-computes chunk-level key-value (KV) caches, (ii) stitches them together at inference time using independent–attention and reordered-RoPE techniques, and (iii) preserves answer quality without changing the model architecture. Hence, online computation of KV caches is eliminated during inference. Our approach is applicable to most existing large language models and their applications without any requirement in modification of models and inference systems. Experimental results across a suite of RAG benchmarks demonstrate that TurboRAG reduces TTFT by up to 9.4x compared to the conventional RAG systems (on an average of 8.6x), but reserving comparable performance to the standard RAG systems.
%U https://aclanthology.org/2025.emnlp-main.334/
%P 6599-6612
Markdown (Informal)
[TurboRAG: Accelerating Retrieval-Augmented Generation with Precomputed KV Caches for Chunked Text](https://aclanthology.org/2025.emnlp-main.334/) (Lu et al., EMNLP 2025)
ACL