@inproceedings{yu-etal-2026-format,
title = "Format as Nuisance: Intervention-Inspired Cross-Format Distillation for Knowledge Conflicts",
author = "Yu, Yue and
Luo, Wang and
Ye, Shenghao and
Bao, Zepeng and
Liu, Jiacheng and
Li, Yongqi and
Tang, Wenxin and
Qian, Tieyun",
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.2028/",
pages = "46190--46211",
ISBN = "979-8-89176-459-0",
abstract = "Large Language Models (LLMs) encode parametric knowledge but can exhibit format bias when resolving knowledge conflicts. When evidence is rendered as text, tables, or knowledge graphs, LLMs can favor certain formats despite unchanged semantic content. Existing work mainly diagnoses this bias, while existing inference-time mitigation remains tied to a single rendered prompt. We propose FaDe, a cross-format knowledge distillation framework that treats evidence format as a controllable nuisance variable and marginalizes teacher predictions over controlled format assignments. FaDe constructs a format-marginalized soft target from these renderings, then trains the student with an asymmetric two-path objective that combines heterogeneous-format debiasing with all-text anchoring. Experiments on ConflictBank and two QA-derived controlled conflict sets with four model backbones show FaDe reduces format bias while improving dual-source coverage and cross-format stability under single-pass inference. Our code is available at https://github.com/Kaia91/FaDe."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yu-etal-2026-format">
<titleInfo>
<title>Format as Nuisance: Intervention-Inspired Cross-Format Distillation for Knowledge Conflicts</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wang</namePart>
<namePart type="family">Luo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shenghao</namePart>
<namePart type="family">Ye</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zepeng</namePart>
<namePart type="family">Bao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiacheng</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yongqi</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wenxin</namePart>
<namePart type="family">Tang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tieyun</namePart>
<namePart type="family">Qian</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>Large Language Models (LLMs) encode parametric knowledge but can exhibit format bias when resolving knowledge conflicts. When evidence is rendered as text, tables, or knowledge graphs, LLMs can favor certain formats despite unchanged semantic content. Existing work mainly diagnoses this bias, while existing inference-time mitigation remains tied to a single rendered prompt. We propose FaDe, a cross-format knowledge distillation framework that treats evidence format as a controllable nuisance variable and marginalizes teacher predictions over controlled format assignments. FaDe constructs a format-marginalized soft target from these renderings, then trains the student with an asymmetric two-path objective that combines heterogeneous-format debiasing with all-text anchoring. Experiments on ConflictBank and two QA-derived controlled conflict sets with four model backbones show FaDe reduces format bias while improving dual-source coverage and cross-format stability under single-pass inference. Our code is available at https://github.com/Kaia91/FaDe.</abstract>
<identifier type="citekey">yu-etal-2026-format</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.2028/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>46190</start>
<end>46211</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Format as Nuisance: Intervention-Inspired Cross-Format Distillation for Knowledge Conflicts
%A Yu, Yue
%A Luo, Wang
%A Ye, Shenghao
%A Bao, Zepeng
%A Liu, Jiacheng
%A Li, Yongqi
%A Tang, Wenxin
%A Qian, Tieyun
%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 yu-etal-2026-format
%X Large Language Models (LLMs) encode parametric knowledge but can exhibit format bias when resolving knowledge conflicts. When evidence is rendered as text, tables, or knowledge graphs, LLMs can favor certain formats despite unchanged semantic content. Existing work mainly diagnoses this bias, while existing inference-time mitigation remains tied to a single rendered prompt. We propose FaDe, a cross-format knowledge distillation framework that treats evidence format as a controllable nuisance variable and marginalizes teacher predictions over controlled format assignments. FaDe constructs a format-marginalized soft target from these renderings, then trains the student with an asymmetric two-path objective that combines heterogeneous-format debiasing with all-text anchoring. Experiments on ConflictBank and two QA-derived controlled conflict sets with four model backbones show FaDe reduces format bias while improving dual-source coverage and cross-format stability under single-pass inference. Our code is available at https://github.com/Kaia91/FaDe.
%U https://aclanthology.org/2026.emnlp-main.2028/
%P 46190-46211
Markdown (Informal)
[Format as Nuisance: Intervention-Inspired Cross-Format Distillation for Knowledge Conflicts](https://aclanthology.org/2026.emnlp-main.2028/) (Yu et al., EMNLP 2026)
ACL
- Yue Yu, Wang Luo, Shenghao Ye, Zepeng Bao, Jiacheng Liu, Yongqi Li, Wenxin Tang, and Tieyun Qian. 2026. Format as Nuisance: Intervention-Inspired Cross-Format Distillation for Knowledge Conflicts. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 46190–46211, Budapest, Hungary. Association for Computational Linguistics.