afrimgsm_common_yaml

group:
 - mgsm_direct
 - afrimgsm
dataset_path: masakhane/afrimgsm
output_type: generate_until
training_split: train
test_split: test
fewshot_split: train
target_delimiter: ""
doc_to_target: '{% if answer is not none %}{{answer}}{% else %}{{answer_number|string}}{% endif %}'
doc_to_text: '{% if answer is not none %}{{question+"\nAnswer:"}}{% else %}{{"Question: "+question+"\nAnswer:"}}{% endif %}'
generation_kwargs:
  do_sample: false
  until:
  - 'Question:'
  - </s>
  - <|im_end|>
filter_list:
  - filter:
    - function: regex
      group_select: -1
      regex_pattern: (-?[0-9.,]{2,})|(-?[0-9]+)
    - function: take_first
    name: flexible-extract
  - filter:
    - function: regex-numbers
      group_select: -1
      regex_pattern: (\d{1,10}(?:,\d{3})*(?:[.,]\d{3})?)([^\d()]*)
    - function: take_first
    name: flexible-extract-new
metric_list:
  - metric: exact_match
    aggregation: mean
    higher_is_better: true
    ignore_case: true
    ignore_punctuation: true
  - metric: f1
    aggregation: !function utils.weighted_f1_score
    average: weighted
    higher_is_better: True
    ignore_case: true
    ignore_punctuation: true
  - metric: squad
    aggregation: squad_f1
    average: weighted
    higher_is_better: True
    ignore_case: true
    ignore_punctuation: true
metadata:
  version: 2.0