@inproceedings{jiang-luo-2026-chart,
title = "Chart-{EARL}: Improving Chart Reasoning of Vision {LLM}s via Early Anchoring in Reinforcement Learning",
author = "Jiang, Gongyao and
Luo, Qiong",
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.1591/",
doi = "10.18653/2026.findings-emnlp.1591",
pages = "33872--33889",
ISBN = "979-8-89176-462-0",
abstract = "Chart reasoning poses a significant challenge for Vision Large Language Models (VLLMs). While Reinforcement Learning (RL) has emerged as a promising solution, existing outcome-based RL methods overlook intermediate errors, whereas process-based approaches remain difficult in verifying abstract logical steps. We therefore analyze chart-reasoning failures to identify an intermediate supervision signal that is both relevant to the task and amenable to objective verification. Our pilot study suggests that early visual-perception errors constitute a prominent source of failure in the chart-reasoning setting. Guided by this insight, we propose Chart-EARL, which constructs verifiable early visual anchors for targeted chart-reasoning supervision. We first build a structured training corpus that explicitly decouples visual perception from logical deduction. Building upon this structure, our approach optimizes the model with rewards for both accurate extraction of descriptive visual features and final-answer correctness. This strategy provides direct and verifiable feedback on early visual grounding while keeping supervision on the final reasoning objective. Experiments show clear performance improvements on complex chart reasoning benchmarks."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jiang-luo-2026-chart">
<titleInfo>
<title>Chart-EARL: Improving Chart Reasoning of Vision LLMs via Early Anchoring in Reinforcement Learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Gongyao</namePart>
<namePart type="family">Jiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qiong</namePart>
<namePart type="family">Luo</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>Chart reasoning poses a significant challenge for Vision Large Language Models (VLLMs). While Reinforcement Learning (RL) has emerged as a promising solution, existing outcome-based RL methods overlook intermediate errors, whereas process-based approaches remain difficult in verifying abstract logical steps. We therefore analyze chart-reasoning failures to identify an intermediate supervision signal that is both relevant to the task and amenable to objective verification. Our pilot study suggests that early visual-perception errors constitute a prominent source of failure in the chart-reasoning setting. Guided by this insight, we propose Chart-EARL, which constructs verifiable early visual anchors for targeted chart-reasoning supervision. We first build a structured training corpus that explicitly decouples visual perception from logical deduction. Building upon this structure, our approach optimizes the model with rewards for both accurate extraction of descriptive visual features and final-answer correctness. This strategy provides direct and verifiable feedback on early visual grounding while keeping supervision on the final reasoning objective. Experiments show clear performance improvements on complex chart reasoning benchmarks.</abstract>
<identifier type="citekey">jiang-luo-2026-chart</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.1591</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1591/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>33872</start>
<end>33889</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Chart-EARL: Improving Chart Reasoning of Vision LLMs via Early Anchoring in Reinforcement Learning
%A Jiang, Gongyao
%A Luo, Qiong
%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 jiang-luo-2026-chart
%X Chart reasoning poses a significant challenge for Vision Large Language Models (VLLMs). While Reinforcement Learning (RL) has emerged as a promising solution, existing outcome-based RL methods overlook intermediate errors, whereas process-based approaches remain difficult in verifying abstract logical steps. We therefore analyze chart-reasoning failures to identify an intermediate supervision signal that is both relevant to the task and amenable to objective verification. Our pilot study suggests that early visual-perception errors constitute a prominent source of failure in the chart-reasoning setting. Guided by this insight, we propose Chart-EARL, which constructs verifiable early visual anchors for targeted chart-reasoning supervision. We first build a structured training corpus that explicitly decouples visual perception from logical deduction. Building upon this structure, our approach optimizes the model with rewards for both accurate extraction of descriptive visual features and final-answer correctness. This strategy provides direct and verifiable feedback on early visual grounding while keeping supervision on the final reasoning objective. Experiments show clear performance improvements on complex chart reasoning benchmarks.
%R 10.18653/2026.findings-emnlp.1591
%U https://aclanthology.org/2026.findings-emnlp.1591/
%U https://doi.org/10.18653/2026.findings-emnlp.1591
%P 33872-33889
Markdown (Informal)
[Chart-EARL: Improving Chart Reasoning of Vision LLMs via Early Anchoring in Reinforcement Learning](https://aclanthology.org/2026.findings-emnlp.1591/) (Jiang & Luo, Findings 2026)
ACL