数据集:
lukaemon/mmlu
评估大规模多任务语言理解: https://github.com/hendrycks/test
@article{hendryckstest2021, title={Measuring Massive Multitask Language Understanding}, author={Dan Hendrycks and Collin Burns and Steven Basart and Andy Zou and Mantas Mazeika and Dawn Song and Jacob Steinhardt}, journal={Proceedings of the International Conference on Learning Representations (ICLR)}, year={2021} }