metadata
dataset_info:
features:
- name: prompt
list:
- name: content
dtype: string
- name: role
dtype: string
- name: chosen
list:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected
list:
- name: content
dtype: string
- name: role
dtype: string
- name: ended_thought
dtype: bool
splits:
- name: train
num_bytes: 66784035
num_examples: 19154
download_size: 33426348
dataset_size: 66784035
configs:
- config_name: default
data_files:
- split: train
path: data/train-*