metadata
dataset_info:
features:
- name: ins
dtype: string
- name: pref
dtype: string
- name: outputs
sequence: string
- name: rank
sequence: int32
splits:
- name: train
num_bytes: 117201
num_examples: 50
download_size: 78524
dataset_size: 117201
configs:
- config_name: default
data_files:
- split: train
path: data/train-*