japan-ai-official/jifeval

JIF-Eval是一个日语指令跟随评估数据集，旨在测试语言模型遵循特定日语指令的能力。它基于IFEval框架，并针对日语评估进行了调整。数据集包含了多种类型的指令约束，用于评估、基准测试、研究和微调日语指令跟随能力。JIF-Eval is a Japanese instruction-following evaluation dataset designed to test language models ability to follow specific instructions in Japanese. It is based on the IFEval framework but adapted for Japanese language evaluation. The dataset includes various types of instruction constraints for evaluating, benchmarking, researching, and fine-tuning Japanese instruction-following capabilities.