@inproceedings{dialameh-etal-2025-echo,
title = "{ECHO}-{LL}a{MA}: Efficient Caching for High-Performance {LL}a{MA} Training",
author = "Dialameh, Maryam and
Karim, Rezaul and
Rajabzadeh, Hossein and
Mohamed Awad, Omar and
Chen, Boxing and
Kwon, Hyock Ju and
Ahmed, Walid and
Liu, Yang",
editor = "Potdar, Saloni and
Rojas-Barahona, Lina and
Montella, Sebastien",
booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: Industry Track",
month = nov,
year = "2025",
address = "Suzhou (China)",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.emnlp-industry.156/",
pages = "2252--2269",
ISBN = "979-8-89176-333-3",
abstract = "This paper introduces ECHO-LLaMA, an efficient LLaMA architecture designed to improve both the training speed and inference throughput of LLaMA architectures while maintaining its learning capacity. ECHO-LLaMA transforms LLaMA models into shared KV caching across certain layers, significantly reducing KV computational complexity while maintaining or improving language performance. Experimental results demonstrate that ECHO-LLaMA achieves up to 77{\%} higher token-per-second throughput during training, up to 16{\%} higher Model FLOPs Utilization (MFU), and up to 14{\%} lower loss when trained on an equal number of tokens. Furthermore, on the 1.1B model, ECHO-LLaMA delivers approximately 7{\%} higher test-time throughput compared to the baseline. By introducing a computationally efficient adaptation mechanism, ECHO-LLaMA offers a scalable and cost-effective solution for pretraining and finetuning large language models, enabling faster and more resource-efficient training without compromising performance."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="dialameh-etal-2025-echo">
<titleInfo>
<title>ECHO-LLaMA: Efficient Caching for High-Performance LLaMA Training</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maryam</namePart>
<namePart type="family">Dialameh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rezaul</namePart>
<namePart type="family">Karim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hossein</namePart>
<namePart type="family">Rajabzadeh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Omar</namePart>
<namePart type="family">Mohamed Awad</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Boxing</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hyock</namePart>
<namePart type="given">Ju</namePart>
<namePart type="family">Kwon</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Walid</namePart>
<namePart type="family">Ahmed</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yang</namePart>
<namePart type="family">Liu</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: Industry Track</title>
</titleInfo>
<name type="personal">
<namePart type="given">Saloni</namePart>
<namePart type="family">Potdar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lina</namePart>
<namePart type="family">Rojas-Barahona</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sebastien</namePart>
<namePart type="family">Montella</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">979-8-89176-333-3</identifier>
</relatedItem>
<abstract>This paper introduces ECHO-LLaMA, an efficient LLaMA architecture designed to improve both the training speed and inference throughput of LLaMA architectures while maintaining its learning capacity. ECHO-LLaMA transforms LLaMA models into shared KV caching across certain layers, significantly reducing KV computational complexity while maintaining or improving language performance. Experimental results demonstrate that ECHO-LLaMA achieves up to 77% higher token-per-second throughput during training, up to 16% higher Model FLOPs Utilization (MFU), and up to 14% lower loss when trained on an equal number of tokens. Furthermore, on the 1.1B model, ECHO-LLaMA delivers approximately 7% higher test-time throughput compared to the baseline. By introducing a computationally efficient adaptation mechanism, ECHO-LLaMA offers a scalable and cost-effective solution for pretraining and finetuning large language models, enabling faster and more resource-efficient training without compromising performance.</abstract>
<identifier type="citekey">dialameh-etal-2025-echo</identifier>
<location>
<url>https://aclanthology.org/2025.emnlp-industry.156/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>2252</start>
<end>2269</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ECHO-LLaMA: Efficient Caching for High-Performance LLaMA Training
%A Dialameh, Maryam
%A Karim, Rezaul
%A Rajabzadeh, Hossein
%A Mohamed Awad, Omar
%A Chen, Boxing
%A Kwon, Hyock Ju
%A Ahmed, Walid
%A Liu, Yang
%Y Potdar, Saloni
%Y Rojas-Barahona, Lina
%Y Montella, Sebastien
%S Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: Industry Track
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou (China)
%@ 979-8-89176-333-3
%F dialameh-etal-2025-echo
%X This paper introduces ECHO-LLaMA, an efficient LLaMA architecture designed to improve both the training speed and inference throughput of LLaMA architectures while maintaining its learning capacity. ECHO-LLaMA transforms LLaMA models into shared KV caching across certain layers, significantly reducing KV computational complexity while maintaining or improving language performance. Experimental results demonstrate that ECHO-LLaMA achieves up to 77% higher token-per-second throughput during training, up to 16% higher Model FLOPs Utilization (MFU), and up to 14% lower loss when trained on an equal number of tokens. Furthermore, on the 1.1B model, ECHO-LLaMA delivers approximately 7% higher test-time throughput compared to the baseline. By introducing a computationally efficient adaptation mechanism, ECHO-LLaMA offers a scalable and cost-effective solution for pretraining and finetuning large language models, enabling faster and more resource-efficient training without compromising performance.
%U https://aclanthology.org/2025.emnlp-industry.156/
%P 2252-2269
Markdown (Informal)
[ECHO-LLaMA: Efficient Caching for High-Performance LLaMA Training](https://aclanthology.org/2025.emnlp-industry.156/) (Dialameh et al., EMNLP 2025)
ACL
- Maryam Dialameh, Rezaul Karim, Hossein Rajabzadeh, Omar Mohamed Awad, Boxing Chen, Hyock Ju Kwon, Walid Ahmed, and Yang Liu. 2025. ECHO-LLaMA: Efficient Caching for High-Performance LLaMA Training. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 2252–2269, Suzhou (China). Association for Computational Linguistics.