metadata
dataset_info:
features:
- name: inputs_embeds
sequence: float64
- name: rewards
dtype: float64
splits:
- name: train
num_bytes: 702999880
num_examples: 21446
- name: validation
num_bytes: 175766360
num_examples: 5362
download_size: 208720652
dataset_size: 878766240
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*