---
dataset_info:
  features:
  - name: uid
    dtype: string
  - name: image
    dtype: image
  - name: caption
    dtype: string
  - name: detections
    list:
    - name: label
      dtype: string
    - name: bbox
      sequence: float32
      length: 4
    - name: confidence
      dtype: float32
  splits:
  - name: train
    num_bytes: 2941330675.0
    num_examples: 72158
  download_size: 2897856826
  dataset_size: 2941330675.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---