@inproceedings{oh-oh-2025-beyond,
title = "Beyond Fixed-Length Calibration for Post-Training Compression of {LLM}s",
author = "Oh, Jaehoon and
Oh, Dokwan",
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.1054/",
doi = "10.18653/v1/2025.findings-emnlp.1054",
pages = "19355--19366",
ISBN = "979-8-89176-335-7",
abstract = "As large language models (LLMs) continue to grow in size, their practical deployment increasingly relies on a range of compression techniques, such as quantization, pruning, and low-rank approximation. Especially, post-training compression methods{--}which do not require re-training{--}have drawn considerable interest. Many recent methods leverage calibration data to capture magnitude or second-order characteristics of input activations. However, the role and significance of calibration data remain underexplored. In this study, we demonstrate that the sequence length of calibration data plays a crucial role in the effectiveness of post-training compression methods for LLMs. We then analyze input activations and find that, within the normalized hidden states, the embedding of the first token exhibits characteristics opposite to those of subsequent tokens. Building on this insight, we introduce state-aware length calibration, a technique that applies masking along the sequence axis, specifically targeting normalized hidden states. Experimental results show that our approach improves perplexity and zero-shot downstream tasks performance."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="oh-oh-2025-beyond">
<titleInfo>
<title>Beyond Fixed-Length Calibration for Post-Training Compression of LLMs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jaehoon</namePart>
<namePart type="family">Oh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dokwan</namePart>
<namePart type="family">Oh</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>As large language models (LLMs) continue to grow in size, their practical deployment increasingly relies on a range of compression techniques, such as quantization, pruning, and low-rank approximation. Especially, post-training compression methods–which do not require re-training–have drawn considerable interest. Many recent methods leverage calibration data to capture magnitude or second-order characteristics of input activations. However, the role and significance of calibration data remain underexplored. In this study, we demonstrate that the sequence length of calibration data plays a crucial role in the effectiveness of post-training compression methods for LLMs. We then analyze input activations and find that, within the normalized hidden states, the embedding of the first token exhibits characteristics opposite to those of subsequent tokens. Building on this insight, we introduce state-aware length calibration, a technique that applies masking along the sequence axis, specifically targeting normalized hidden states. Experimental results show that our approach improves perplexity and zero-shot downstream tasks performance.</abstract>
<identifier type="citekey">oh-oh-2025-beyond</identifier>
<identifier type="doi">10.18653/v1/2025.findings-emnlp.1054</identifier>
<location>
<url>https://aclanthology.org/2025.findings-emnlp.1054/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>19355</start>
<end>19366</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Beyond Fixed-Length Calibration for Post-Training Compression of LLMs
%A Oh, Jaehoon
%A Oh, Dokwan
%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 oh-oh-2025-beyond
%X As large language models (LLMs) continue to grow in size, their practical deployment increasingly relies on a range of compression techniques, such as quantization, pruning, and low-rank approximation. Especially, post-training compression methods–which do not require re-training–have drawn considerable interest. Many recent methods leverage calibration data to capture magnitude or second-order characteristics of input activations. However, the role and significance of calibration data remain underexplored. In this study, we demonstrate that the sequence length of calibration data plays a crucial role in the effectiveness of post-training compression methods for LLMs. We then analyze input activations and find that, within the normalized hidden states, the embedding of the first token exhibits characteristics opposite to those of subsequent tokens. Building on this insight, we introduce state-aware length calibration, a technique that applies masking along the sequence axis, specifically targeting normalized hidden states. Experimental results show that our approach improves perplexity and zero-shot downstream tasks performance.
%R 10.18653/v1/2025.findings-emnlp.1054
%U https://aclanthology.org/2025.findings-emnlp.1054/
%U https://doi.org/10.18653/v1/2025.findings-emnlp.1054
%P 19355-19366
Markdown (Informal)
[Beyond Fixed-Length Calibration for Post-Training Compression of LLMs](https://aclanthology.org/2025.findings-emnlp.1054/) (Oh & Oh, Findings 2025)
ACL