@inproceedings{cao-etal-2026-scaling,
title = "Scaling Laws for Mixed Quantization",
author = "Cao, Zeyu and
Lai, Yao and
Gu, Boyang and
Zhang, Cheng and
Gimenes, Pedro and
Lu, Jianqiao and
Gao, Xitong and
Cheng, Jianyi and
Mullins, Robert D. and
Zhao, Yiren",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.142/",
pages = "3265--3308",
ISBN = "979-8-89176-459-0",
abstract = "Post-training quantization (PTQ) of large language models (LLMs) reduces memory usage and accelerates inference while maintaining accuracy. However, it remains unclear how to select a quantization scheme that achieves target accuracy across different model scales. We introduce two parameters to characterize this: the quantization ratio $Q_r$ (proportion of parameters quantized to low precision) and the block size $Q_b$ (number of parameters sharing a scaling factor). Based on extensive experiments across multiple LLM families and scales, we propose a unified scaling law that predicts quantization-induced loss degradation as a function of $Q_r$, $Q_b$, and model size. Our scaling law reveals two practical implications: (1) larger models tolerate higher $Q_r$, making mixed-precision increasingly effective at scale; (2) small $Q_b$, commonly used in mixed precision, becomes unnecessary for large models, simplifying hardware design without sacrificing accuracy. These findings provide a principled recipe for quantizing LLMs in practical deployment."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="cao-etal-2026-scaling">
<titleInfo>
<title>Scaling Laws for Mixed Quantization</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zeyu</namePart>
<namePart type="family">Cao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yao</namePart>
<namePart type="family">Lai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Boyang</namePart>
<namePart type="family">Gu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Cheng</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pedro</namePart>
<namePart type="family">Gimenes</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jianqiao</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xitong</namePart>
<namePart type="family">Gao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jianyi</namePart>
<namePart type="family">Cheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Robert</namePart>
<namePart type="given">D</namePart>
<namePart type="family">Mullins</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yiren</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Post-training quantization (PTQ) of large language models (LLMs) reduces memory usage and accelerates inference while maintaining accuracy. However, it remains unclear how to select a quantization scheme that achieves target accuracy across different model scales. We introduce two parameters to characterize this: the quantization ratio Q_r (proportion of parameters quantized to low precision) and the block size Q_b (number of parameters sharing a scaling factor). Based on extensive experiments across multiple LLM families and scales, we propose a unified scaling law that predicts quantization-induced loss degradation as a function of Q_r, Q_b, and model size. Our scaling law reveals two practical implications: (1) larger models tolerate higher Q_r, making mixed-precision increasingly effective at scale; (2) small Q_b, commonly used in mixed precision, becomes unnecessary for large models, simplifying hardware design without sacrificing accuracy. These findings provide a principled recipe for quantizing LLMs in practical deployment.</abstract>
<identifier type="citekey">cao-etal-2026-scaling</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.142/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>3265</start>
<end>3308</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Scaling Laws for Mixed Quantization
%A Cao, Zeyu
%A Lai, Yao
%A Gu, Boyang
%A Zhang, Cheng
%A Gimenes, Pedro
%A Lu, Jianqiao
%A Gao, Xitong
%A Cheng, Jianyi
%A Mullins, Robert D.
%A Zhao, Yiren
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F cao-etal-2026-scaling
%X Post-training quantization (PTQ) of large language models (LLMs) reduces memory usage and accelerates inference while maintaining accuracy. However, it remains unclear how to select a quantization scheme that achieves target accuracy across different model scales. We introduce two parameters to characterize this: the quantization ratio Q_r (proportion of parameters quantized to low precision) and the block size Q_b (number of parameters sharing a scaling factor). Based on extensive experiments across multiple LLM families and scales, we propose a unified scaling law that predicts quantization-induced loss degradation as a function of Q_r, Q_b, and model size. Our scaling law reveals two practical implications: (1) larger models tolerate higher Q_r, making mixed-precision increasingly effective at scale; (2) small Q_b, commonly used in mixed precision, becomes unnecessary for large models, simplifying hardware design without sacrificing accuracy. These findings provide a principled recipe for quantizing LLMs in practical deployment.
%U https://aclanthology.org/2026.emnlp-main.142/
%P 3265-3308
Markdown (Informal)
[Scaling Laws for Mixed Quantization](https://aclanthology.org/2026.emnlp-main.142/) (Cao et al., EMNLP 2026)
ACL
- Zeyu Cao, Yao Lai, Boyang Gu, Cheng Zhang, Pedro Gimenes, Jianqiao Lu, Xitong Gao, Jianyi Cheng, Robert D. Mullins, and Yiren Zhao. 2026. Scaling Laws for Mixed Quantization. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 3265–3308, Budapest, Hungary. Association for Computational Linguistics.