metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: pixel_values
sequence:
sequence:
sequence:
sequence: float64
- name: image_sizes
sequence: int64
splits:
- name: train
num_bytes: 83148021162
num_examples: 1802
download_size: 13708443932
dataset_size: 83148021162
configs:
- config_name: default
data_files:
- split: train
path: data/train-*