dataset_info:
features:
- name: prompt
dtype: string - name: chosen
dtype: string - name: rejected
dtype: string - name: prompt_length
dtype: int64 - name: chosen_length
dtype: int64 - name: rejected_length
dtype: int64 - name: source_dataset
dtype: string
splits: - name: train
num_bytes: 82619579.0
num_examples: 43228 - name: test
num_bytes: 839284.0
num_examples: 439
download_size: 45993474
dataset_size: 83458863.0
configs: - config_name: default
data_files:- split: train
path: data/train-* - split: test
path: data/test-*
- split: train
A combination of the following preference datasets