metadata
dataset_info:
features:
- name: conversation
list:
- name: content
dtype: string
- name: role
dtype: string
- name: chosen
struct:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected
struct:
- name: content
dtype: string
- name: role
dtype: string
- name: output
dtype: string
- name: input
dtype: string
- name: instruction
dtype: string
splits:
- name: train
num_bytes: 2085664
num_examples: 856
download_size: 1102241
dataset_size: 2085664
configs:
- config_name: default
data_files:
- split: train
path: data/train-*