metadata
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: float32
- name: pixel_mask
sequence:
sequence: int64
- name: labels
sequence: float32
splits:
- name: validation
num_bytes: 9840363136
num_examples: 1328
download_size: 319687102
dataset_size: 9840363136
configs:
- config_name: default
data_files:
- split: validation
path: data/validation-*