metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: answers
sequence: string
- name: label
sequence: int64
- name: step_scores
sequence: float64
splits:
- name: train
num_bytes: 767766990
num_examples: 1319
download_size: 145318010
dataset_size: 767766990
configs:
- config_name: default
data_files:
- split: train
path: data/train-*