Dataset repo requires arbitrary Python code execution

[Dylannnnnnnn]

May I ask the reason for this: gaia-benchmark/GAIA · Datasets at Hugging Face

I can assess it just now. This is a popular dataset in large language models. Can anyone help me?