group: bbh dataset_path: lukaemon/bbh output_type: greedy_until test_split: test doc_to_text: "{{input}}" doc_to_target: "{{target}}" metric_list: - metric: exact_match aggregation: mean higher_is_better: true ignore_case: true ignore_punctuation: false generation_kwargs: until: - "\n\n" do_sample: false temperature: 0.0