introvoyz041/ChemO

ChemO数据集是基于2025年国际化学奥林匹克竞赛（IChO）构建的基准数据集，代表了自动化化学问题解决的新前沿。该数据集包含来自IChO 2025的9个问题，每个问题都以结构化的JSON文件形式提供（1.json ~ 9.json在`JSON/`目录中）。数据集的特点包括：奥林匹克级别的挑战性问题、多模态符号语言（涵盖文本、公式和分子结构）、以及两种新颖的评估方法（AER和SVE）。数据集旨在用于多模态大型语言模型（MLLM）基准测试、多智能体系统测试和多模态推理。数据源明确指向ICHO 2025的官方网站。当前版本未包含CDXML文件，但未来会更新补充。

The ChemO dataset is a benchmark built from the International Chemistry Olympiad (IChO) 2025, representing a new frontier in automated chemical problem-solving. It includes 9 problems from IChO 2025, each provided as a structured JSON file (1.json ~ 9.json in `JSON/`). Key features of the dataset include: Olympic-level challenging problems, multimodal symbolic language (covering text, formulas, and molecular structures), and two novel assessment methods (AER and SVE). The dataset is designed for Multimodal Large Language Model (MLLM) benchmarking, multi-agent system testing, and multimodal reasoning. The data source is clearly linked to the official IChO 2025 website. The current release does not include CDXML files but plans to supplement them in future updates.