---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: metadata
    struct:
    - name: id
      dtype: int64
    - name: image
      dtype: string
    - name: width
      dtype: int64
    - name: height
      dtype: int64
    - name: conversations
      list:
      - name: from
        dtype: string
      - name: value
        dtype: string
  splits:
  - name: train
    num_bytes: 7373827894.9
    num_examples: 79900
  - name: validation
    num_bytes: 955654176.16
    num_examples: 10480
  - name: test
    num_bytes: 876866031.54
    num_examples: 9620
  download_size: 8937554691
  dataset_size: 9206348102.599998
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
---