metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: chosen
dtype: string
- name: rejected
dtype: string
- name: chosen_model
dtype: int64
- name: prompt_tokens
dtype: int64
- name: completion_tokens
dtype: int64
- name: total_tokens
dtype: int64
- name: error_msg
dtype: string
splits:
- name: train
num_bytes: 656786
num_examples: 520
download_size: 327207
dataset_size: 656786
configs:
- config_name: default
data_files:
- split: train
path: data/train-*