metadata
dataset_info:
config_name: zh_tw
features:
- name: messages
list:
- name: content
dtype: string
- name: image
dtype: string
- name: role
dtype: string
- name: image
dtype: image
- name: assistant
dtype: string
- name: score
dtype: float64
- name: eval-seed
dtype: float64
- name: conversation_id
dtype: string
splits:
- name: train
num_bytes: 4236657297.624
num_examples: 25768
download_size: 4094881357
dataset_size: 4236657297.624
configs:
- config_name: zh_tw
data_files:
- split: train
path: zh_tw/train-*