PyFi: Toward Pyramid-like Financial Image Understanding for VLMs via Adversarial Agents
This work addresses the challenge of financial image understanding for VLMs, though it is incremental as it builds on existing VLM capabilities with a new dataset and fine-tuning approach.
The paper tackles the problem of enabling vision language models (VLMs) to reason through complex financial images by proposing PyFi, a framework that uses a pyramid-structured dataset of 600K synthesized question-answer pairs to fine-tune models, resulting in average accuracy improvements of 19.52% and 8.06% for specific VLMs.
This paper proposes PyFi, a novel framework for pyramid-like financial image understanding that enables vision language models (VLMs) to reason through question chains in a progressive, simple-to-complex manner. At the core of PyFi is PyFi-600K, a dataset comprising 600K financial question-answer pairs organized into a reasoning pyramid: questions at the base require only basic perception, while those toward the apex demand increasing levels of capability in financial visual understanding and expertise. This data is scalable because it is synthesized without human annotations, using PyFi-adv, a multi-agent adversarial mechanism under the Monte Carlo Tree Search (MCTS) paradigm, in which, for each image, a challenger agent competes with a solver agent by generating question chains that progressively probe deeper capability levels in financial visual reasoning. Leveraging this dataset, we present fine-grained, hierarchical, and comprehensive evaluations of advanced VLMs in the financial domain. Moreover, fine-tuning Qwen2.5-VL-3B and Qwen2.5-VL-7B on the pyramid-structured question chains enables these models to answer complex financial questions by decomposing them into sub-questions with gradually increasing reasoning demands, yielding average accuracy improvements of 19.52% and 8.06%, respectively, on the dataset. All resources of code, dataset and models are available at: https://github.com/AgenticFinLab/PyFi .