metadata
dataset_info:
features:
- name: chosen
dtype: string
- name: rejected
dtype: string
- name: parsed_chosen
list:
- name: content
dtype: string
- name: role
dtype: string
- name: parsed_rejected
list:
- name: content
dtype: string
- name: role
dtype: string
- name: tokenized_and_annotated_chosen
sequence: int64
- name: tokenized_and_annotated_rejected
sequence: int64
splits:
- name: train
num_bytes: 276365744
num_examples: 60043
download_size: 132350690
dataset_size: 276365744
configs:
- config_name: default
data_files:
- split: train
path: data/train-*