---
dataset_info:
  features:
  - name: id
    dtype: int32
  - name: image
    dtype: image
  - name: description
    dtype: string
  - name: conversations
    list:
    - name: role
      dtype: string
    - name: content
      dtype: string
  splits:
  - name: train
    num_bytes: 27615612831.52
    num_examples: 205346
  download_size: 28102458218
  dataset_size: 27615612831.52
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

Next 200,000 images with more than 1,200,000 QnA pairs are coming soon !