@inproceedings{liu-etal-2026-one,
title = "One Forward Beats Two: {I}nner{Z}oom for Accurate and Efficient {GUI} Grounding",
author = "Liu, Chen and
Chen, Ling and
Zhou, Hanzhang and
Chen, Liangyu and
Cai, Chenglin and
Yu, Xin and
Hoi, Steven and
Wang, Yue",
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.843/",
pages = "19560--19581",
ISBN = "979-8-89176-459-0",
abstract = "MLLM-based GUI grounding methods commonly formulate target localization as autoregressive coordinate generation, enabling models to leverage strong instruction following and semantic understanding. However, by casting localization as coordinate-token prediction, these methods rely on implicit decoding to convert region-level target evidence into precise click coordinates, making point-level spatial alignment unreliable. Through diagnostic analysis, we find that intermediate decoder layers already exhibit target-region awareness, yet such region-level evidence fails to guide final point-level decisions. To address this, we propose InnerZoom, a single-forward cross-layer evidence bridging framework that refines target evidence inside the model during decoding, avoiding extra inferences required by Zoom-In. InnerZoom reframes coordinate decoding as an evidence-preserving process by identifying target-related cues within the original forward pass, refining them across decoder layers, and reinjecting the refined evidence into coordinate prediction. Extensive experiments show that InnerZoom-4B achieves state-of-the-art performance on all six mainstream GUI grounding benchmarks, outperforming the second-best result by 4.1 points on OSWorld-G.Under the 4B setting, InnerZoom improves the same SFT+RL baseline by 5.3 points on average and outperforms two-pass Zoom-In by 1.3 points on average, while reducing measured GPU inference latency by 44{\%} and estimated TFLOPs by 29{\%}. Code and models will be at https://github.com/YenanLiu/InnerZoom."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="liu-etal-2026-one">
<titleInfo>
<title>One Forward Beats Two: InnerZoom for Accurate and Efficient GUI Grounding</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chen</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ling</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hanzhang</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Liangyu</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chenglin</namePart>
<namePart type="family">Cai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xin</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Steven</namePart>
<namePart type="family">Hoi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Wang</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>MLLM-based GUI grounding methods commonly formulate target localization as autoregressive coordinate generation, enabling models to leverage strong instruction following and semantic understanding. However, by casting localization as coordinate-token prediction, these methods rely on implicit decoding to convert region-level target evidence into precise click coordinates, making point-level spatial alignment unreliable. Through diagnostic analysis, we find that intermediate decoder layers already exhibit target-region awareness, yet such region-level evidence fails to guide final point-level decisions. To address this, we propose InnerZoom, a single-forward cross-layer evidence bridging framework that refines target evidence inside the model during decoding, avoiding extra inferences required by Zoom-In. InnerZoom reframes coordinate decoding as an evidence-preserving process by identifying target-related cues within the original forward pass, refining them across decoder layers, and reinjecting the refined evidence into coordinate prediction. Extensive experiments show that InnerZoom-4B achieves state-of-the-art performance on all six mainstream GUI grounding benchmarks, outperforming the second-best result by 4.1 points on OSWorld-G.Under the 4B setting, InnerZoom improves the same SFT+RL baseline by 5.3 points on average and outperforms two-pass Zoom-In by 1.3 points on average, while reducing measured GPU inference latency by 44% and estimated TFLOPs by 29%. Code and models will be at https://github.com/YenanLiu/InnerZoom.</abstract>
<identifier type="citekey">liu-etal-2026-one</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.843/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>19560</start>
<end>19581</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T One Forward Beats Two: InnerZoom for Accurate and Efficient GUI Grounding
%A Liu, Chen
%A Chen, Ling
%A Zhou, Hanzhang
%A Chen, Liangyu
%A Cai, Chenglin
%A Yu, Xin
%A Hoi, Steven
%A Wang, Yue
%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 liu-etal-2026-one
%X MLLM-based GUI grounding methods commonly formulate target localization as autoregressive coordinate generation, enabling models to leverage strong instruction following and semantic understanding. However, by casting localization as coordinate-token prediction, these methods rely on implicit decoding to convert region-level target evidence into precise click coordinates, making point-level spatial alignment unreliable. Through diagnostic analysis, we find that intermediate decoder layers already exhibit target-region awareness, yet such region-level evidence fails to guide final point-level decisions. To address this, we propose InnerZoom, a single-forward cross-layer evidence bridging framework that refines target evidence inside the model during decoding, avoiding extra inferences required by Zoom-In. InnerZoom reframes coordinate decoding as an evidence-preserving process by identifying target-related cues within the original forward pass, refining them across decoder layers, and reinjecting the refined evidence into coordinate prediction. Extensive experiments show that InnerZoom-4B achieves state-of-the-art performance on all six mainstream GUI grounding benchmarks, outperforming the second-best result by 4.1 points on OSWorld-G.Under the 4B setting, InnerZoom improves the same SFT+RL baseline by 5.3 points on average and outperforms two-pass Zoom-In by 1.3 points on average, while reducing measured GPU inference latency by 44% and estimated TFLOPs by 29%. Code and models will be at https://github.com/YenanLiu/InnerZoom.
%U https://aclanthology.org/2026.emnlp-main.843/
%P 19560-19581
Markdown (Informal)
[One Forward Beats Two: InnerZoom for Accurate and Efficient GUI Grounding](https://aclanthology.org/2026.emnlp-main.843/) (Liu et al., EMNLP 2026)
ACL
- Chen Liu, Ling Chen, Hanzhang Zhou, Liangyu Chen, Chenglin Cai, Xin Yu, Steven Hoi, and Yue Wang. 2026. One Forward Beats Two: InnerZoom for Accurate and Efficient GUI Grounding. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 19560–19581, Budapest, Hungary. Association for Computational Linguistics.