A 700-sample evaluation subset drawn from MMLU-Pro, formatted for LLM evaluation pipelines.
Split
Samples
Subjects
test
700
14
Subjects: math, health, physics, business, biology, chemistry, computer science, economics, engineering, philosophy, other, history, psychology, law.
Column
Type
Description
prompt
list[dict]
Chat-format prompt with few-shot examples (role / content).… See the full description on the dataset page:
https://huggingface.co/datasets/SeanWang0027/mmlu_pro_subset_test.