metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: chosen
dtype: string
- name: rejected
dtype: string
splits:
- name: train
num_bytes: 37257823
num_examples: 25000
- name: test
num_bytes: 36475685
num_examples: 25000
- name: unsupervised
num_bytes: 74756794
num_examples: 50000
download_size: 88255132
dataset_size: 148490302
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
- split: unsupervised
path: data/unsupervised-*
tags:
- RLHF
- DPO
- preferences
IMDB dataset formated as preference pairs