@inproceedings{wang-etal-2026-tta,
title = "{AT}-{TTA}: Dual-View Anchor-Tethered Test-Time Adaptation for Large Language Models",
author = "Wang, Yaxin and
He, Junfeng and
Lu, Zhihui and
Lu, Weishen and
Hu, Shijing and
Chen, Peng and
Wang, Ye and
Huang, Chao and
Cheng, Yao and
Cheng, Yudong",
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.1374/",
pages = "31353--31381",
ISBN = "979-8-89176-459-0",
abstract = "Large Language Models (LLMs) frequently exhibit performance degradation in specialized domains due to distribution shifts. While unsupervised Test-Time Adaptation (TTA) offers a lightweight solution, existing methods suffer from critical flaws: input-centric adaptation lacks reasoning constraints, while output-centric optimization falls into a ``Confidence Trap'' of reinforced hallucinations. We instead study Few-Shot Anchored TTA, where a small labeled support set is available, and propose Anchor-Tethered Test-Time Adaptation (AT-TTA), a framework that synergizes linguistic alignment with reliable reasoning. AT-TTA employs Dual-View Uncertainty Calibration, minimizing input perplexity for domain alignment and sequence-level output energy for generation consistency. Furthermore, to mitigate catastrophic drift, we construct a ``Semantic Anchor'' by fine-tuning on the labeled support set and its filtered augmentations, which serves as a stable domain prior and constrains the adaptation trajectory via an Energy-Guided Adaptive Replay strategy. Extensive experiments on the AdaptEval benchmark demonstrate strong performance across diverse domains, effectively preventing model collapse while maintaining high inference efficiency."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2026-tta">
<titleInfo>
<title>AT-TTA: Dual-View Anchor-Tethered Test-Time Adaptation for Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yaxin</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junfeng</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhihui</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weishen</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shijing</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Peng</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ye</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chao</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yao</namePart>
<namePart type="family">Cheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yudong</namePart>
<namePart type="family">Cheng</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) frequently exhibit performance degradation in specialized domains due to distribution shifts. While unsupervised Test-Time Adaptation (TTA) offers a lightweight solution, existing methods suffer from critical flaws: input-centric adaptation lacks reasoning constraints, while output-centric optimization falls into a “Confidence Trap” of reinforced hallucinations. We instead study Few-Shot Anchored TTA, where a small labeled support set is available, and propose Anchor-Tethered Test-Time Adaptation (AT-TTA), a framework that synergizes linguistic alignment with reliable reasoning. AT-TTA employs Dual-View Uncertainty Calibration, minimizing input perplexity for domain alignment and sequence-level output energy for generation consistency. Furthermore, to mitigate catastrophic drift, we construct a “Semantic Anchor” by fine-tuning on the labeled support set and its filtered augmentations, which serves as a stable domain prior and constrains the adaptation trajectory via an Energy-Guided Adaptive Replay strategy. Extensive experiments on the AdaptEval benchmark demonstrate strong performance across diverse domains, effectively preventing model collapse while maintaining high inference efficiency.</abstract>
<identifier type="citekey">wang-etal-2026-tta</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1374/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>31353</start>
<end>31381</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T AT-TTA: Dual-View Anchor-Tethered Test-Time Adaptation for Large Language Models
%A Wang, Yaxin
%A He, Junfeng
%A Lu, Zhihui
%A Lu, Weishen
%A Hu, Shijing
%A Chen, Peng
%A Wang, Ye
%A Huang, Chao
%A Cheng, Yao
%A Cheng, Yudong
%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 wang-etal-2026-tta
%X Large Language Models (LLMs) frequently exhibit performance degradation in specialized domains due to distribution shifts. While unsupervised Test-Time Adaptation (TTA) offers a lightweight solution, existing methods suffer from critical flaws: input-centric adaptation lacks reasoning constraints, while output-centric optimization falls into a “Confidence Trap” of reinforced hallucinations. We instead study Few-Shot Anchored TTA, where a small labeled support set is available, and propose Anchor-Tethered Test-Time Adaptation (AT-TTA), a framework that synergizes linguistic alignment with reliable reasoning. AT-TTA employs Dual-View Uncertainty Calibration, minimizing input perplexity for domain alignment and sequence-level output energy for generation consistency. Furthermore, to mitigate catastrophic drift, we construct a “Semantic Anchor” by fine-tuning on the labeled support set and its filtered augmentations, which serves as a stable domain prior and constrains the adaptation trajectory via an Energy-Guided Adaptive Replay strategy. Extensive experiments on the AdaptEval benchmark demonstrate strong performance across diverse domains, effectively preventing model collapse while maintaining high inference efficiency.
%U https://aclanthology.org/2026.emnlp-main.1374/
%P 31353-31381
Markdown (Informal)
[AT-TTA: Dual-View Anchor-Tethered Test-Time Adaptation for Large Language Models](https://aclanthology.org/2026.emnlp-main.1374/) (Wang et al., EMNLP 2026)
ACL
- Yaxin Wang, Junfeng He, Zhihui Lu, Weishen Lu, Shijing Hu, Peng Chen, Ye Wang, Chao Huang, Yao Cheng, and Yudong Cheng. 2026. AT-TTA: Dual-View Anchor-Tethered Test-Time Adaptation for Large Language Models. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 31353–31381, Budapest, Hungary. Association for Computational Linguistics.