metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: prompt_id
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: score_chosen
dtype: float64
- name: score_rejected
dtype: float64
splits:
- name: train
num_bytes: 13188881.497490898
num_examples: 2507
download_size: 2896405
dataset_size: 13188881.497490898
configs:
- config_name: default
data_files:
- split: train
path: data/train-*