POPE_sample / README.md
Oztobuzz's picture
Initial upload of synthetic OKVQA dataset with key objects
bb747f7 verified
|
raw
history blame
541 Bytes
metadata
dataset_info:
  features:
    - name: idx
      dtype: float64
    - name: img
      dtype: image
    - name: original_question
      dtype: string
    - name: gt_ans
      dtype: string
    - name: key_object
      dtype: string
    - name: question
      dtype: string
    - name: answer
      dtype: string
    - name: vqa_sample
      dtype: string
  splits:
    - name: train
      num_bytes: 203829
      num_examples: 5
  download_size: 85081
  dataset_size: 203829
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/default/train-*