Shiyunee/HonestyBench

HonestyBench是一个大规模的问题回答基准数据集，它整合了10个流行的公开自由形式事实性问题回答数据集，包含了560k的训练样本，以及38k的域内和33k的域外评估样本。该数据集旨在帮助模型在多种任务上达到性能上限，并为不同方法的比较提供了一个健壮可靠的测试平台。HonestyBench is a large-scale question-answering benchmark that consolidates 10 widely used public freeform factual question-answering datasets, containing 560k training samples, and 38k in-domain and 33k out-of-domain (OOD) evaluation samples. It is designed to help models achieve the upper bound of performance across diverse tasks and serves as a robust and reliable testbed for comparing different approaches.