Japanese commonsense reasoning QA dataset with 5-choice questions.
pip install lm-eval && lm_eval --model hf --tasks jcommonsenseqa