metadata
dataset_info:
features:
- name: input_ids
sequence:
sequence: int32
- name: attention_mask
sequence: int8
- name: ppl
dtype: float64
splits:
- name: train
num_bytes: 1772753580
num_examples: 172783
download_size: 594832297
dataset_size: 1772753580
configs:
- config_name: default
data_files:
- split: train
path: data/train-*