@inproceedings{nguyen-etal-2026-complementary,
title = "Complementary Lattice Correction for Post-Training Quantization",
author = "Nguyen, Tung and
Tran, Quang Hung and
Van, Linh Ngo and
Hoang, Dinh Thai and
Nguyen, Diep and
Le, Trung and
Nguyen, Duc Anh",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.158/",
pages = "3309--3334",
ISBN = "979-8-89176-462-0",
abstract = "Weight-only post-training quantization (PTQ) for large language models (LLMs) typically ends with the projection of floating-point weights onto a uniform integer lattice, introducing structured output error due to non-zero activation means. Existing bias correction methods compensate for this shift by adding bias terms, which is incompatible with modern bias-less architectures such as LLaMA and Mistral and only addresses the error indirectly. We propose **Complementary Lattice Correction (CLC)**, a lightweight post-quantization method that corrects projection error directly on the integer lattice without modifying the model architecture. CLC performs a budget-constrained greedy descent that selectively flips quantized weights to adjacent lattice levels. We further show that CLC provably reduces the first-moment error and admits a worst-case bound on the layer-wise mean-squared error change, which certifies non-increase under a verifiable budget condition. Experiments across multiple models, quantizers, and bit widths show that CLC consistently improves perplexity and downstream accuracy with negligible overhead."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="nguyen-etal-2026-complementary">
<titleInfo>
<title>Complementary Lattice Correction for Post-Training Quantization</title>
</titleInfo>
<name type="personal">
<namePart type="given">Tung</namePart>
<namePart type="family">Nguyen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Quang</namePart>
<namePart type="given">Hung</namePart>
<namePart type="family">Tran</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Linh</namePart>
<namePart type="given">Ngo</namePart>
<namePart type="family">Van</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dinh</namePart>
<namePart type="given">Thai</namePart>
<namePart type="family">Hoang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Diep</namePart>
<namePart type="family">Nguyen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Trung</namePart>
<namePart type="family">Le</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Duc</namePart>
<namePart type="given">Anh</namePart>
<namePart type="family">Nguyen</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Weight-only post-training quantization (PTQ) for large language models (LLMs) typically ends with the projection of floating-point weights onto a uniform integer lattice, introducing structured output error due to non-zero activation means. Existing bias correction methods compensate for this shift by adding bias terms, which is incompatible with modern bias-less architectures such as LLaMA and Mistral and only addresses the error indirectly. We propose **Complementary Lattice Correction (CLC)**, a lightweight post-quantization method that corrects projection error directly on the integer lattice without modifying the model architecture. CLC performs a budget-constrained greedy descent that selectively flips quantized weights to adjacent lattice levels. We further show that CLC provably reduces the first-moment error and admits a worst-case bound on the layer-wise mean-squared error change, which certifies non-increase under a verifiable budget condition. Experiments across multiple models, quantizers, and bit widths show that CLC consistently improves perplexity and downstream accuracy with negligible overhead.</abstract>
<identifier type="citekey">nguyen-etal-2026-complementary</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.158/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>3309</start>
<end>3334</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Complementary Lattice Correction for Post-Training Quantization
%A Nguyen, Tung
%A Tran, Quang Hung
%A Van, Linh Ngo
%A Hoang, Dinh Thai
%A Nguyen, Diep
%A Le, Trung
%A Nguyen, Duc Anh
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F nguyen-etal-2026-complementary
%X Weight-only post-training quantization (PTQ) for large language models (LLMs) typically ends with the projection of floating-point weights onto a uniform integer lattice, introducing structured output error due to non-zero activation means. Existing bias correction methods compensate for this shift by adding bias terms, which is incompatible with modern bias-less architectures such as LLaMA and Mistral and only addresses the error indirectly. We propose **Complementary Lattice Correction (CLC)**, a lightweight post-quantization method that corrects projection error directly on the integer lattice without modifying the model architecture. CLC performs a budget-constrained greedy descent that selectively flips quantized weights to adjacent lattice levels. We further show that CLC provably reduces the first-moment error and admits a worst-case bound on the layer-wise mean-squared error change, which certifies non-increase under a verifiable budget condition. Experiments across multiple models, quantizers, and bit widths show that CLC consistently improves perplexity and downstream accuracy with negligible overhead.
%U https://aclanthology.org/2026.findings-emnlp.158/
%P 3309-3334
Markdown (Informal)
[Complementary Lattice Correction for Post-Training Quantization](https://aclanthology.org/2026.findings-emnlp.158/) (Nguyen et al., Findings 2026)
ACL
- Tung Nguyen, Quang Hung Tran, Linh Ngo Van, Dinh Thai Hoang, Diep Nguyen, Trung Le, and Duc Anh Nguyen. 2026. Complementary Lattice Correction for Post-Training Quantization. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 3309–3334, Budapest, Hungary. Association for Computational Linguistics.