lm_eval/tasks/sciq/sciq.yaml

task: sciq
dataset_path: sciq
dataset_name: null
output_type: multiple_choice
training_split: train
validation_split: validation
test_split: test
doc_to_text: "{{support.lstrip()}}\nQuestion: {{question}}\nAnswer:"
doc_to_target: 3
doc_to_choice: "{{[distractor1, distractor2, distractor3, correct_answer]}}"
should_decontaminate: true
doc_to_decontamination_query: "{{support}} {{question}}"
metric_list:
  - metric: acc
    aggregation: mean
    higher_is_better: true
  - metric: acc_norm
    aggregation: mean
    higher_is_better: true
metadata:
  version: 1.0