Search papers, labs, and topics across Lattice.
The paper introduces HierVA, a hierarchical visual agent framework designed to improve multi-step reasoning across multiple subplots in chart question answering. HierVA employs a high-level manager to generate plans and maintain a compact context, while specialized workers perform reasoning and gather evidence, using separate visual and textual contexts. Experiments on the CharXiv dataset show that HierVA outperforms existing MLLMs, demonstrating the benefits of hierarchical architecture, scoped visual context, and distilled context.
A hierarchical agent that separates visual and textual contexts drastically improves multi-step reasoning on complex charts, outperforming monolithic MLLMs.
Advanced chart question answering requires both precise perception of small visual elements and multi-step reasoning across several subplots. While existing MLLMs are strong at understanding single plots, they often struggle with multi-step reasoning across multiple subplots. We propose HierVA, a hierarchical visual agent framework for chart reasoning that iteratively constructs and updates a working context in a joint image--text space. A high-level manager generates plans and maintains a compact context containing only key information, while specialized workers perform reasoning, gather evidence, and return results. In particular, the agent maintains separate visual and textual contexts, using a zoom-in tool to restrict the visual context. Experiments on the CharXiv reasoning subset demonstrate consistent improvements over strong multimodal baselines, and ablation studies verify that hierarchical architecture, scoped visual context, and distilled context contribute complementary gains.
