FINANCEBENCH
收藏资源简介:
FINANCEBENCH是由斯坦福大学等机构创建的金融问答测试套件,包含10,231个关于公开交易公司的问题,每个问题都有相应的答案和证据字符串。数据集旨在评估大型语言模型在开放式金融问答任务中的性能。数据集覆盖了多样化的场景,旨在作为最低性能标准。数据集的创建过程涉及多学科专家团队,包括AI、评估和金融服务领域的专家。FINANCEBENCH的应用领域主要集中在金融分析和投资决策支持,旨在解决金融领域特定任务的模型评估问题。
FINANCEBENCH is a financial question answering test suite developed by Stanford University and other institutions. It contains 10,231 questions about publicly traded companies, with each question paired with a corresponding answer and evidence string. This dataset is designed to evaluate the performance of large language models (LLMs) on open-ended financial question answering tasks. It covers diverse scenarios and is intended to serve as a minimum performance benchmark. The creation of FINANCEBENCH involved a multidisciplinary team of experts spanning the fields of artificial intelligence, evaluation methodology, and financial services. The primary application areas of FINANCEBENCH focus on financial analysis and investment decision support, with the goal of addressing model evaluation challenges for specific tasks in the financial domain.

- 1FinanceBench: A New Benchmark for Financial Question Answering斯坦福大学 · 2023年



