metadata
dataset_info:
features:
- name: prompt
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: category
struct:
- name: animal_abuse
dtype: bool
- name: child_abuse
dtype: bool
- name: controversial_topics,politics
dtype: bool
- name: discrimination,stereotype,injustice
dtype: bool
- name: drug_abuse,weapons,banned_substance
dtype: bool
- name: financial_crime,property_crime,theft
dtype: bool
- name: hate_speech,offensive_language
dtype: bool
- name: misinformation_regarding_ethics,laws_and_safety
dtype: bool
- name: non_violent_unethical_behavior
dtype: bool
- name: privacy_violation
dtype: bool
- name: self_harm
dtype: bool
- name: sexually_explicit,adult_content
dtype: bool
- name: terrorism,organized_crime
dtype: bool
- name: violence,aiding_and_abetting,incitement
dtype: bool
splits:
- name: train
num_bytes: 37675784
num_examples: 10000
download_size: 16124540
dataset_size: 37675784
configs:
- config_name: default
data_files:
- split: train
path: data/train-*