metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: ground_truth_chosen
dtype: string
- name: v1_rejected
dtype: string
- name: v1_1_rejected
dtype: string
splits:
- name: train
num_bytes: 6026820159
num_examples: 2154856
- name: validation
num_bytes: 319232514.11141086
num_examples: 114548
download_size: 3316539077
dataset_size: 6346052673.111411
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*