OK_VQA_sample / README.md
Oztobuzz's picture
Initial upload of synthetic OKVQA dataset with key objects
5a795fb verified
metadata
license: mit
configs:
  - config_name: Try_2
    data_files:
      - split: train
        path: data/Try_2/train-*
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
dataset_info:
  - config_name: Try_2
    features:
      - name: idx
        dtype: int64
      - name: img
        dtype: image
      - name: question
        dtype: string
      - name: gt_ans
        sequence: string
      - name: key_object
        dtype: string
      - name: question_detection
        dtype: string
      - name: question_type
        dtype: string
    splits:
      - name: train
        num_bytes: 5295313
        num_examples: 100
    download_size: 5278305
    dataset_size: 5295313
  - config_name: default
    features:
      - name: idx
        dtype: int64
      - name: img
        dtype: image
      - name: question
        dtype: string
      - name: gt_ans
        sequence: string
      - name: key_object
        dtype: string
      - name: question_detection
        dtype: string
      - name: question_type
        dtype: string
    splits:
      - name: train
        num_bytes: 5295347
        num_examples: 100
    download_size: 5278435
    dataset_size: 5295347