@inproceedings{yang-etal-2026-targeted,
title = "When Targeted Grounded Gains Hide Complementary Abstraction Regression: {DION}-Bench and a Conjunctive Evaluation Protocol",
author = "Yang, Zida and
You, Guquan and
Zheng, Jiexin and
Peng, Yijie",
editor = "Wang, Yimu and
Choi, Yee Man and
Wu, Di and
Ouyang, Siqi and
Azadani, Mozhgan Nasr and
Fung, Yi R. (May)",
booktitle = "Proceedings of the 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently ({G}round{LM} 2026)",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.groundlm-1.77/",
pages = "1091--1115",
ISBN = "979-8-89176-478-1",
abstract = "Post-training with internal feedback can improve a targeted knowledge-grounded behavior while silently degrading complementary abstraction and logical-consistency capabilities. We propose a conjunctive evaluation protocol that accepts a method only when a designated target improves and complementary probes remain non-inferior (with a stricter secondary criterion requiring complementary equivalence), while also reporting retention and predictive drift. We instantiate the protocol in DION-Bench, combining two lexical/ontological probes (same-category exceptions; abstraction recovery) with a synthetic SAT-checked logical-consistency probe. On Qwen3-8B-Base, a task-specific self-judged DPO adaptation improves OntoNeg-WN by 5.27 percentage points but reduces ConceptShift by 37.50 points{---}despite humans reaching 82.7{\%} and frontier LLMs 44{--}73{\%} on OntoNeg{---}whereas mean change across five external accuracy suites is only {\ensuremath{-}}1.70 points. A development-tuned decoding rule partially mitigates ConceptShift regression to 12.93/14.60 points on the original/holdout splits while retaining most of the target gain. These results show that faithful post-training claims require joint reporting of gain, complementary abstraction regression, retention, drift, and equivalence{---}not proof that a single shared knowledge item degraded."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yang-etal-2026-targeted">
<titleInfo>
<title>When Targeted Grounded Gains Hide Complementary Abstraction Regression: DION-Bench and a Conjunctive Evaluation Protocol</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zida</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guquan</namePart>
<namePart type="family">You</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiexin</namePart>
<namePart type="family">Zheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yijie</namePart>
<namePart type="family">Peng</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 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently (GroundLM 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yimu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yee</namePart>
<namePart type="given">Man</namePart>
<namePart type="family">Choi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Di</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siqi</namePart>
<namePart type="family">Ouyang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mozhgan</namePart>
<namePart type="given">Nasr</namePart>
<namePart type="family">Azadani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yi</namePart>
<namePart type="given">R</namePart>
<namePart type="given">(May)</namePart>
<namePart type="family">Fung</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">979-8-89176-478-1</identifier>
</relatedItem>
<abstract>Post-training with internal feedback can improve a targeted knowledge-grounded behavior while silently degrading complementary abstraction and logical-consistency capabilities. We propose a conjunctive evaluation protocol that accepts a method only when a designated target improves and complementary probes remain non-inferior (with a stricter secondary criterion requiring complementary equivalence), while also reporting retention and predictive drift. We instantiate the protocol in DION-Bench, combining two lexical/ontological probes (same-category exceptions; abstraction recovery) with a synthetic SAT-checked logical-consistency probe. On Qwen3-8B-Base, a task-specific self-judged DPO adaptation improves OntoNeg-WN by 5.27 percentage points but reduces ConceptShift by 37.50 points—despite humans reaching 82.7% and frontier LLMs 44–73% on OntoNeg—whereas mean change across five external accuracy suites is only \ensuremath-1.70 points. A development-tuned decoding rule partially mitigates ConceptShift regression to 12.93/14.60 points on the original/holdout splits while retaining most of the target gain. These results show that faithful post-training claims require joint reporting of gain, complementary abstraction regression, retention, drift, and equivalence—not proof that a single shared knowledge item degraded.</abstract>
<identifier type="citekey">yang-etal-2026-targeted</identifier>
<location>
<url>https://aclanthology.org/2026.groundlm-1.77/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>1091</start>
<end>1115</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T When Targeted Grounded Gains Hide Complementary Abstraction Regression: DION-Bench and a Conjunctive Evaluation Protocol
%A Yang, Zida
%A You, Guquan
%A Zheng, Jiexin
%A Peng, Yijie
%Y Wang, Yimu
%Y Choi, Yee Man
%Y Wu, Di
%Y Ouyang, Siqi
%Y Azadani, Mozhgan Nasr
%Y Fung, Yi R. (May)
%S Proceedings of the 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently (GroundLM 2026)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-478-1
%F yang-etal-2026-targeted
%X Post-training with internal feedback can improve a targeted knowledge-grounded behavior while silently degrading complementary abstraction and logical-consistency capabilities. We propose a conjunctive evaluation protocol that accepts a method only when a designated target improves and complementary probes remain non-inferior (with a stricter secondary criterion requiring complementary equivalence), while also reporting retention and predictive drift. We instantiate the protocol in DION-Bench, combining two lexical/ontological probes (same-category exceptions; abstraction recovery) with a synthetic SAT-checked logical-consistency probe. On Qwen3-8B-Base, a task-specific self-judged DPO adaptation improves OntoNeg-WN by 5.27 percentage points but reduces ConceptShift by 37.50 points—despite humans reaching 82.7% and frontier LLMs 44–73% on OntoNeg—whereas mean change across five external accuracy suites is only \ensuremath-1.70 points. A development-tuned decoding rule partially mitigates ConceptShift regression to 12.93/14.60 points on the original/holdout splits while retaining most of the target gain. These results show that faithful post-training claims require joint reporting of gain, complementary abstraction regression, retention, drift, and equivalence—not proof that a single shared knowledge item degraded.
%U https://aclanthology.org/2026.groundlm-1.77/
%P 1091-1115
Markdown (Informal)
[When Targeted Grounded Gains Hide Complementary Abstraction Regression: DION-Bench and a Conjunctive Evaluation Protocol](https://aclanthology.org/2026.groundlm-1.77/) (Yang et al., GroundLM 2026)
ACL