README.md

metadata

dataset_info:
  features:
    - name: image
      dtype: image
    - name: filename
      dtype: string
    - name: imgid
      dtype: int32
    - name: sentids
      list: int32
    - name: sentences
      list:
        - name: tokens
          list: string
        - name: raw
          dtype: string
        - name: imgid
          dtype: int32
        - name: sentid
          dtype: int32
    - name: raw_turkish
      sequence: string
  splits:
    - name: train
      num_bytes: 3843262999
      num_examples: 29000
    - name: validation
      num_bytes: 141460570.428
      num_examples: 1014
    - name: test
      num_bytes: 143011292
      num_examples: 1000
  download_size: 4318471070
  dataset_size: 4127734861.428
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*