dataset_path: Helsinki-NLP/tatoeba_mt
training_split: validation
test_split: test
output_type: generate_until
dataset_kwargs:
  trust_remote_code: true
metric_list:
  - metric: bleu
    higher_is_better: true
  - metric: chrf
    higher_is_better: true
generation_kwargs:
  until:
    - "\n"
  do_sample: false
  num_beams: 1
  max_new_tokens: 256
metadata:
  version: 1.0
