FinNLP-FNP-LLMFinLegal-2025 Regulation Challenge Evaluation Data
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/Open-Finance-Lab/Regulations_Challenge_COLING_2025
下载链接
链接失效反馈官方服务：
资源简介：
该数据集用于评估RKEFino1模型在基于知识和数学推理的问答任务上的表现。它包含了准确性、FactScore以及F1分数等性能评估指标。在规模上，数据集包含了150个针对知识问答（K-QA）的测试样本，以及50个针对数学推理问答（MR-QA）的测试样本。任务的范畴涵盖了基于知识的问答（K-QA）和数学推理问答（MR-QA）。
This dataset is used to evaluate the performance of the RKEFino1 model on knowledge-based and mathematical reasoning-based question answering tasks. It includes performance evaluation metrics such as accuracy, FactScore, and F1-score. In terms of scale, the dataset contains 150 test samples for knowledge-based question answering (K-QA) and 50 test samples for mathematical reasoning-based question answering (MR-QA). The task scope covers both knowledge-based question answering (K-QA) and mathematical reasoning-based question answering (MR-QA).
提供机构：
Open Finance Lab
