FinMR
收藏资源简介:
FinMR是一个高质量、知识密集的多模态数据集,专为评估专业分析师级别的金融推理能力而设计。该数据集包含超过3200个精心策划和专家注释的问题-答案对,涵盖15个不同的金融主题,确保了广泛的领域多样性,并在多种图像类型中整合了复杂的数学推理、先进的金融知识和微妙的视觉解释任务。通过使用丰富的视觉内容和详细的解释注释,FinMR为评估和推进多模态金融推理向专业分析师级别的竞争力提供了一个重要的基准工具。
FinMR is a high-quality, knowledge-intensive multimodal dataset specifically designed to evaluate financial reasoning capabilities at the professional analyst level. This dataset includes over 3,200 carefully curated and expert-annotated question-answer pairs, covering 15 distinct financial topics to ensure broad domain diversity, and integrates complex mathematical reasoning, advanced financial knowledge and subtle visual interpretation tasks across multiple image types. By utilizing rich visual content and detailed explanatory annotations, FinMR serves as a critical benchmark tool for evaluating and advancing multimodal financial reasoning to attain professional analyst-level competitiveness.

- 1FinMR: A Knowledge-Intensive Multimodal Benchmark for Advanced Financial Reasoning奥克兰大学(新西兰) · 2025年



