metadata
dataset_info:
features:
- name: instruction
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: reason
dtype: string
splits:
- name: train
num_bytes: 90684561
num_examples: 19000
- name: test
num_bytes: 2513063
num_examples: 523
download_size: 27420469
dataset_size: 93197624
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*