lm_eval/tasks/tinyBenchmarks/tinyMMLU.yaml

task: tinyMMLU
dataset_path: tinyBenchmarks/tinyMMLU
dataset_name: all
test_split: test
fewshot_split: dev
num_fewshot: 0
fewshot_config:
  sampler: first_n
output_type: multiple_choice
doc_to_text: "{{input_formatted}}"
doc_to_choice: ["A", "B", "C", "D"]
doc_to_target: answer
num_fewshot: 0
metric_list:
  - metric: acc_norm
    aggregation: !function agg_functions.agg_gpirt_mmlu
    higher_is_better: true
metadata:
  version: 0.0