dyck_language_task / README.md
Jeffrey
Upload dataset
b74ba5d verified
metadata
dataset_info:
  features:
    - name: env
      dtype: string
    - name: prompt
      dtype: string
    - name: extra
      struct:
        - name: game_data
          struct:
            - name: answer
              dtype: string
            - name: difficulty
              dtype: int64
            - name: gpt_response
              dtype: string
            - name: metadata
              struct:
                - name: closing_sequence
                  dtype: string
                - name: fill_length
                  dtype: int64
                - name: full_sequence
                  dtype: string
                - name: n_types
                  dtype: int64
                - name: nesting_depth
                  dtype: int64
                - name: question_sequence
                  dtype: string
                - name: seed
                  dtype: int64
                - name: total_length
                  dtype: int64
                - name: trace_id
                  dtype: string
            - name: question
              dtype: string
        - name: metadata
          struct:
            - name: closing_sequence
              dtype: string
            - name: fill_length
              dtype: int64
            - name: full_sequence
              dtype: string
            - name: n_types
              dtype: int64
            - name: nesting_depth
              dtype: int64
            - name: question_sequence
              dtype: string
            - name: seed
              dtype: int64
            - name: total_length
              dtype: int64
            - name: trace_id
              dtype: string
        - name: seed
          dtype: int64
        - name: task_id
          dtype: int64
        - name: task_type
          dtype: string
  splits:
    - name: train
      num_bytes: 5762898
      num_examples: 5093
  download_size: 4216720
  dataset_size: 5762898
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*