story_cloze.yaml

# The name of the dataset to use (via the datasets library).
dataset_name: &path 'story_cloze'

# The configuration name of the dataset to use (via the datasets library).
dataset_config_name: &name '2016'

# The split name of the dataset to use (via the datasets library).
split: &split 'test'


# Specify the arguments required to utilize datasets.load_dataset
# https://huggingface.co/docs/datasets/package_reference/loading_methods#datasets.load_dataset
DATASET_KWARGS:
  path: *path
  name: *name
  data_dir: "./extraction/story_cloze"
  split: *split

# Arguments for proper data preprocessing
DATASET_INFO: 
  # Number of classes
  num_classes: 2
  # For dynamic task, specify column names for candidate answers
  choices: ['sentence_quiz1', 'sentence_quiz2']
  # Correct Answer Label Column Name
  label: 'answer_right_ending'
  # Dynamic task or not
  is_dynamic: true

# Arguments required to add a custom prompt
TEMPLATE_INFO:
  # Name of the text column in the dataset to use
  text_formats: ["{{input_sentence_1}}", "{{input_sentence_2}}", "{{input_sentence_3}}", "{{input_sentence_4}}"]
  
  # Refer to the following PromptSource template jinja and modify jinja_suffix.
  # ./extraction/promptsource/templates/{dataset_name/dataset_config_name}/templates.yaml
  jinja_suffix: " ||| {{ answer_choices[answer_right_ending | int() -1] }}"