metadata
dataset_info:
features:
- name: from
dtype: string
- 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
splits:
- name: train
num_bytes: 56746477
num_examples: 33449
download_size: 29970425
dataset_size: 56746477
configs:
- config_name: default
data_files:
- split: train
path: data/train-*