metadata
dataset_info:
features:
- name: id
dtype: string
- name: image
dtype: string
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
- name: conversation
list:
- name: input
dtype: string
- name: need_eos_token
dtype: bool
- name: output
dtype: string
- name: sep
dtype: string
- name: input_ids
sequence: int32
- name: labels
sequence: int64
- name: length
dtype: int64
splits:
- name: train
num_bytes: 458824446
num_examples: 558128
download_size: 109368654
dataset_size: 458824446
configs:
- config_name: default
data_files:
- split: train
path: data/train-*