mgsm_direct_eu.yaml

task: mgsm_direct_eu
dataset_path: HiTZ/MGSM-eu
dataset_name: null
doc_to_target: '{{answer_number|string}}'
doc_to_text: '{% if answer is not none %}{{question+"\nErantzuna:"}}{% else %}{{"Galdera: "+question+"\nErantzuna:"}}{% endif %}'
output_type: generate_until
training_split: train
test_split: test
target_delimiter: " "
generation_kwargs:
  until:
    - "\n\n"
    - "\n"
    - "Galdera:"
    - </s>
    - <|im_end|>
  do_sample: false
  temperature: 0.0
filter_list:
  - name: remove_whitespace
    filter:
      - function: remove_whitespace
      - function: take_first
  - name: flexible-extract
    filter:
    - function: regex
      group_select: -1
      regex_pattern: (-?[0-9]+([ .,][0-9.,]+)?)
    - function: take_first
metric_list:
  - metric: exact_match
    aggregation: mean
    higher_is_better: true
    ignore_case: true
    ignore_punctuation: true
    regexes_to_ignore:
      - " "
metadata:
  version: 1.0