overfit-brothers/KRX-MCQA

该数据集是为第三届KRX金融语言模型竞赛准备的MCQA（多选问题）数据集，由overfit-brothers团队提供。数据集包括多个配置，如example-mcqa、account_calc、financial_market_augmented、financial_mmlu_augmented、krx-textbook和mmmlu_revised_augmented_cot等。这些数据集涵盖了金融领域的多个方面，包括会计原理、金融市场法规、韩国交易所的教育材料等。数据集的生成方式包括直接制作、通过GPT-4 API增强、以及从现有数据集中提取和重构。数据集的语言为韩语，许可证包括CC BY-SA 4.0和MIT许可证。

The dataset consists of multiple sub-datasets, primarily for multiple-choice question-answering tasks in the financial domain. Specifically:
- example-mcqa: Example data from KRX-Bench.
- account_calc: An MCQA calculation problem dataset based on 37 seed data enhanced by gpt-4o API, totaling 222 data entries.
- financial_market_augmented: A multiple-choice dataset in the financial domain, with CoT-formatted explanations generated by gpt-4o API, totaling 1063 data entries.
- financial_mmlu_augmented: Similar to financial_market_augmented but with different sources and license information.
- krx-textbook: A dataset of questions extracted from an e-book published by the Korea Exchange, totaling 212 data entries.
- mmmlu_revised_augmented_cot series: Extracted Korean-language portion from the openai/MMMLU dataset, with different formats of answers generated by gpt-4o API, totaling 6986 data entries.