tag:
  - longbench_e
task: longbench_samsum_e
dataset_path: THUDM/LongBench
test_split: test
dataset_name: samsum_e
doc_to_text: "Summarize the dialogue into a few short sentences. The following are some examples.\n\n{{context}}\n\n{{input}}"
doc_to_target: '{{answers}}'
process_results: !function metrics.get_rouge_score
generation_kwargs:
  max_gen_toks: 128
  temperature: 1
  do_sample: False
  until: ["\n"]
metric_list:
  - metric: "rouge_score"
    aggregation: mean
    higher_is_better: True
metadata:
  version: 4.0