metadata
dataset_info:
features:
- name: model_a
dtype: string
- name: model_b
dtype: string
- name: prompt
list: string
- name: response_a
list: string
- name: conversation_a
list:
- name: role
dtype: string
- name: text
dtype: string
- name: response_b
list: string
- name: conversation_b
list:
- name: role
dtype: string
- name: text
dtype: string
- name: winner
dtype: string
splits:
- name: train
num_bytes: 2492458259
num_examples: 135634
download_size: 1357697091
dataset_size: 2492458259
configs:
- config_name: default
data_files:
- split: train
path: data/train-*