metadata
dataset_info:
features:
- name: improvement_prompt
list:
- name: content
dtype: string
- name: role
dtype: string
- name: metadata
dtype: string
- name: model_responses__improved_response
sequence: string
- name: improvement_prompt__improved_response__metadata
dtype: string
- name: model_responses__improved_response__metadata
dtype: string
splits:
- name: train
num_bytes: 6717224
num_examples: 570
download_size: 1852216
dataset_size: 6717224
configs:
- config_name: default
data_files:
- split: train
path: data/train-*