metadata
dataset_info:
features:
- name: input_ids
list: int32
- name: attention_mask
list: int8
- name: image_grid_thw
list: int32
- name: original_index
dtype: int32
splits:
- name: train
num_bytes: 499515683
num_examples: 199986
download_size: 75359284
dataset_size: 499515683
configs:
- config_name: default
data_files:
- split: train
path: data/train-*