metadata
dataset_info:
features:
- name: pixel_values
sequence:
sequence:
sequence: float32
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: labels
dtype: int64
splits:
- name: train
num_bytes: 8147050794
num_examples: 5994
- name: test
num_bytes: 7875210594
num_examples: 5794
download_size: 4581320669
dataset_size: 16022261388
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*