---
dataset_info:
  features:
  - name: input_ids
    sequence: int32
  - name: token_type_ids
    sequence: int8
  - name: attention_mask
    sequence: int8
  - name: pixel_values
    sequence:
      sequence:
        sequence: float16
  - name: labels
    sequence: int64
  splits:
  - name: train
    num_bytes: 1792257360
    num_examples: 5820
  - name: val
    num_bytes: 768330260
    num_examples: 2495
  download_size: 1221715033
  dataset_size: 2560587620
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: val
    path: data/val-*
---