Datasets:

Trickxter
/

COCO2017-captions

license: cc-by-4.0
dataset_info:
  features:
    - name: image
      dtype: image
    - name: caption
      dtype: string
    - name: coco_url
      dtype: string
    - name: flickr_url
      dtype: string
    - name: date_captured
      dtype: timestamp[s]
  splits:
    - name: train
      num_bytes: 18241178225.122
      num_examples: 118287
    - name: validation
      num_bytes: 787170448
      num_examples: 5000
  download_size: 20103610208
  dataset_size: 19028348673.122
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
task_categories:
  - image-to-text
language:
  - en