metadata
dataset_info:
features:
- name: images
sequence:
image:
decode: false
- name: audios
sequence: 'null'
- name: videos
sequence: 'null'
- name: data
list:
- name: data
dtype: string
- name: modality
dtype: string
- name: role
dtype: string
splits:
- name: train
num_bytes: 4588374257
num_examples: 200000
download_size: 4297562009
dataset_size: 4588374257
configs:
- config_name: default
data_files:
- split: train
path: data/train-*