metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: item_idx
dtype: int64
- name: step_idx
dtype: int64
splits:
- name: train
num_bytes: 29674285672
num_examples: 22870570
download_size: 3071568602
dataset_size: 29674285672
configs:
- config_name: default
data_files:
- split: train
path: data/train-*