dataset_path: TIGER-Lab/MMLU-Pro # a copy of `cais/leaderboard_mmlu` with no auxiliary_train split
task: leaderboard_mmlu_pro
test_split: test
fewshot_split: validation
fewshot_config:
  sampler: first_n
output_type: multiple_choice
doc_to_text: !function utils.doc_to_text
doc_to_choice: !function utils.doc_to_choice
doc_to_target: answer
metric_list:
  - metric: acc
    aggregation: mean
    higher_is_better: true
num_fewshot: 5
metadata:
  version: 0.1