@inproceedings{zhao-etal-2025-position,
title = "Position {ID}s Matter: An Enhanced Position Layout for Efficient Context Compression in Large Language Models",
author = "Zhao, Runsong and
Liu, Xin and
Liu, Xinyu and
Huang, Pengcheng and
Xiao, Chunyang and
Xiao, Tong and
Zhu, JingBo",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-emnlp.962/",
doi = "10.18653/v1/2025.findings-emnlp.962",
pages = "17715--17734",
ISBN = "979-8-89176-335-7",
abstract = "Using special tokens (e.g., gist, memory, or compressed tokens) to compress context information is a common practice for large language models (LLMs). However, existing approaches often neglect that position encodings inherently induce local inductive biases in models, causing the compression process to ignore holistic contextual dependencies. We propose **Enhanced Position Layout (EPL)**, a simple yet effective method that improves the context compression capability of LLMs by only adjusting position IDs, the numerical identifiers that specify token positions. EPL minimizes the distance between context tokens and their corresponding special tokens and at the same time maintains the sequence order in position IDs between context tokens, special tokens, and the subsequent tokens. Integrating EPL into our best performing context compression model results in 1.9 ROUGE-1 F1 improvement on out-of-domain question answering datasets in average. When extended to multimodal scenarios, EPL brings an average accuracy gain of 2.6 to vision compression LLMs."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhao-etal-2025-position">
<titleInfo>
<title>Position IDs Matter: An Enhanced Position Layout for Efficient Context Compression in Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Runsong</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xin</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinyu</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pengcheng</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chunyang</namePart>
<namePart type="family">Xiao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tong</namePart>
<namePart type="family">Xiao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">JingBo</namePart>
<namePart type="family">Zhu</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>Findings of the Association for Computational Linguistics: EMNLP 2025</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">979-8-89176-335-7</identifier>
</relatedItem>
<abstract>Using special tokens (e.g., gist, memory, or compressed tokens) to compress context information is a common practice for large language models (LLMs). However, existing approaches often neglect that position encodings inherently induce local inductive biases in models, causing the compression process to ignore holistic contextual dependencies. We propose **Enhanced Position Layout (EPL)**, a simple yet effective method that improves the context compression capability of LLMs by only adjusting position IDs, the numerical identifiers that specify token positions. EPL minimizes the distance between context tokens and their corresponding special tokens and at the same time maintains the sequence order in position IDs between context tokens, special tokens, and the subsequent tokens. Integrating EPL into our best performing context compression model results in 1.9 ROUGE-1 F1 improvement on out-of-domain question answering datasets in average. When extended to multimodal scenarios, EPL brings an average accuracy gain of 2.6 to vision compression LLMs.</abstract>
<identifier type="citekey">zhao-etal-2025-position</identifier>
<identifier type="doi">10.18653/v1/2025.findings-emnlp.962</identifier>
<location>
<url>https://aclanthology.org/2025.findings-emnlp.962/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>17715</start>
<end>17734</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Position IDs Matter: An Enhanced Position Layout for Efficient Context Compression in Large Language Models
%A Zhao, Runsong
%A Liu, Xin
%A Liu, Xinyu
%A Huang, Pengcheng
%A Xiao, Chunyang
%A Xiao, Tong
%A Zhu, JingBo
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Findings of the Association for Computational Linguistics: EMNLP 2025
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-335-7
%F zhao-etal-2025-position
%X Using special tokens (e.g., gist, memory, or compressed tokens) to compress context information is a common practice for large language models (LLMs). However, existing approaches often neglect that position encodings inherently induce local inductive biases in models, causing the compression process to ignore holistic contextual dependencies. We propose **Enhanced Position Layout (EPL)**, a simple yet effective method that improves the context compression capability of LLMs by only adjusting position IDs, the numerical identifiers that specify token positions. EPL minimizes the distance between context tokens and their corresponding special tokens and at the same time maintains the sequence order in position IDs between context tokens, special tokens, and the subsequent tokens. Integrating EPL into our best performing context compression model results in 1.9 ROUGE-1 F1 improvement on out-of-domain question answering datasets in average. When extended to multimodal scenarios, EPL brings an average accuracy gain of 2.6 to vision compression LLMs.
%R 10.18653/v1/2025.findings-emnlp.962
%U https://aclanthology.org/2025.findings-emnlp.962/
%U https://doi.org/10.18653/v1/2025.findings-emnlp.962
%P 17715-17734
Markdown (Informal)
[Position IDs Matter: An Enhanced Position Layout for Efficient Context Compression in Large Language Models](https://aclanthology.org/2025.findings-emnlp.962/) (Zhao et al., Findings 2025)
ACL