Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Chart-RL: Generalized Chart Comprehension via Reinforcement Learning with Verifiable Rewards

Xin Zhang, Xingyu Li, Rongguang Wang, Ruizhong Miao, Zheng Wang, Dan Roth, Chenyang Li

TL;DR

This work introduces Chart-RL, an effective reinforcement learning (RL) method that employs mathematically verifiable rewards to enhance chart question answering in VLMs and demonstrates that task difficulty and inherent complexity are more critical than data quantity in RL training.

Abstract

Accurate chart comprehension represents a critical challenge in advancing multimodal learning systems, as extensive information is compressed into structured visual representations. However, existing vision-language models (VLMs) frequently struggle to generalize on unseen charts because it requires abstract, symbolic, and quantitative reasoning over structured visual representations. In this work, we introduce Chart-RL, an effective reinforcement learning (RL) method that employs mathematically verifiable rewards to enhance chart question answering in VLMs. Our experiments demonstrate that Chart-RL consistently outperforms supervised fine-tuning (SFT) across different chart understanding benchmarks, achieving relative improvements of 16.7% on MutlChartQA, and 11.5% on ChartInsights. We conduct robustness analysis, where Chart-RL achieves enhanced performance in 18 of 25 perturbed chart categories, demonstrating strong consistency and reasoning capability across visual variations. Furthermore, we demonstrate that task difficulty and inherent complexity are more critical than data quantity in RL training. For instance, Chart-RL trained on merely 10 complex chart-query examples significantly outperforms models trained on over 6,000 simple examples. Additionally, training on challenging reasoning tasks not only improves in-domain generalization relative to simpler tasks, but also facilitate strong transfer to out-of-domain visual mathematical problems.

Chart-RL: Generalized Chart Comprehension via Reinforcement Learning with Verifiable Rewards

TL;DR

This work introduces Chart-RL, an effective reinforcement learning (RL) method that employs mathematically verifiable rewards to enhance chart question answering in VLMs and demonstrates that task difficulty and inherent complexity are more critical than data quantity in RL training.

Abstract

Accurate chart comprehension represents a critical challenge in advancing multimodal learning systems, as extensive information is compressed into structured visual representations. However, existing vision-language models (VLMs) frequently struggle to generalize on unseen charts because it requires abstract, symbolic, and quantitative reasoning over structured visual representations. In this work, we introduce Chart-RL, an effective reinforcement learning (RL) method that employs mathematically verifiable rewards to enhance chart question answering in VLMs. Our experiments demonstrate that Chart-RL consistently outperforms supervised fine-tuning (SFT) across different chart understanding benchmarks, achieving relative improvements of 16.7% on MutlChartQA, and 11.5% on ChartInsights. We conduct robustness analysis, where Chart-RL achieves enhanced performance in 18 of 25 perturbed chart categories, demonstrating strong consistency and reasoning capability across visual variations. Furthermore, we demonstrate that task difficulty and inherent complexity are more critical than data quantity in RL training. For instance, Chart-RL trained on merely 10 complex chart-query examples significantly outperforms models trained on over 6,000 simple examples. Additionally, training on challenging reasoning tasks not only improves in-domain generalization relative to simpler tasks, but also facilitate strong transfer to out-of-domain visual mathematical problems.
Paper Structure (27 sections, 3 equations, 7 figures, 6 tables)

This paper contains 27 sections, 3 equations, 7 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Chart-Based Question Answering
  4. Reinforcement Fine-Tuning in VLMs
  5. Methods
  6. Chart-RL Framework
  7. Accuracy and Format Rewards
  8. Accuracy Reward.
  9. Format Reward.
  10. Experiments
  11. Training & Evaluation Setup
  12. Baseline VLM.
  13. Training Datasets.
  14. Training Hyper-parameters.
  15. Evaluation Benchmarks.
  16. ...and 12 more sections

Figures (7)

  • Figure 1: Comparison of Chart-RL vs SFT to reason on complex chart question answering.
  • Figure 2: Schematic diagram of Chart-RL framework with GRPO optimization for chart comprehension.
  • Figure 3: RL Training reward trajectories for Chart-RL across different training sample sizes, showing accuracy rewards (left) and format rewards (right) throughout the training process.
  • Figure 4: Performance breakdown by sub-category on MultiChartQA and ChartInsights benchmarks.
  • Figure 5: Accuracy reward trajectories in RL training for Easy and Hard tasks.
  • ...and 2 more figures