---
dataset_info:
  features:
  - name: idx
    dtype: int64
  - name: img_url
    dtype: string
  - name: image_tokens
    sequence: int64
  - name: caption
    dtype: string
  - name: clean_caption
    dtype: string
  - name: input_ids
    sequence: int32
  - name: attention_mask
    sequence: int8
  splits:
  - name: train
    num_bytes: 944695861
    num_examples: 527187
  download_size: 174001714
  dataset_size: 944695861
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---