metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: rejected
dtype: string
- name: chosen
dtype: string
- name: pre_score
dtype: float64
- name: post_score
dtype: float64
- name: pre_critique
dtype: string
- name: post_critique
dtype: string
- name: score_diff
dtype: float64
splits:
- name: train_sft
num_bytes: 365604077
num_examples: 77059
- name: test_sft
num_bytes: 75346657
num_examples: 16102
download_size: 210884230
dataset_size: 440950734
configs:
- config_name: default
data_files:
- split: train_sft
path: data/train_sft-*
- split: test_sft
path: data/test_sft-*