tag:
  - longbench
task: longbench_trec
dataset_path: THUDM/LongBench
test_split: test
dataset_name: trec
doc_to_text: 'Please determine the type of the question below. Here are some examples of questions.\n\n{{context}}\n{{input}}'
doc_to_target: '{{answers[0]}}'
process_results: !function metrics.classification_score
generation_kwargs:
  max_gen_toks: 64
  temperature: 1
  do_sample: True
  until: ['\n']
metric_list:
  - metric: "classification_score"
    aggregation: mean
    higher_is_better: True
metadata:
  version: 2.0