metadata
dataset_info:
features:
- name: page_hash
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: normalized_bboxes
sequence:
sequence: int64
- name: labels
sequence: int64
splits:
- name: train
num_bytes: 3793445572
num_examples: 150701
- name: validation
num_bytes: 401719948
num_examples: 15959
- name: test
num_bytes: 275482368
num_examples: 10944
download_size: 110392206
dataset_size: 4470647888
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*