KFinEval-Pilot
收藏资源简介:
KFinEval-Pilot是一个针对韩国金融领域的大型语言模型(LLM)评估的基准套件。由韩国金融电信和清算研究所发起,该数据集包含超过1000个经过精心策划的问题,涵盖金融知识、法律推理和金融毒性三个核心领域,旨在评估LLM在处理韩国金融操作环境中各种现实场景的能力。数据集通过半自动化管道构建,结合GPT-4生成的提示和专家验证,确保领域相关性和事实准确性。
KFinEval-Pilot is a benchmark suite for evaluating large language models (LLMs) in the South Korean financial domain. Initiated by the Korea Financial Telecommunications and Clearings Institute, this dataset includes over 1,000 carefully curated questions covering three core domains: financial knowledge, legal reasoning, and financial toxicity. It aims to assess the capabilities of LLMs in handling various real-world scenarios within the South Korean financial operational environment. The dataset is constructed via a semi-automated pipeline, combining prompts generated by GPT-4 and expert validation to ensure domain relevance and factual accuracy.




