coco2017-captioning / README.md
stephanyresque's picture
Initial upload of COCO 2017 captioning dataset
7b9a3b6 verified
metadata
dataset_info:
  features:
    - name: image
      dtype: image
    - name: image_id
      dtype: int64
    - name: text
      dtype: string
    - name: captions
      list: string
    - name: split
      dtype: string
  splits:
    - name: train
      num_bytes: 19708309805
      num_examples: 118287
    - name: validation
      num_bytes: 802250717
      num_examples: 5000
  download_size: 20172887350
  dataset_size: 20510560522
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*