metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: overflow_to_sample_mapping
dtype: int64
splits:
- name: train
num_bytes: 38586701804
num_examples: 4703401
download_size: 13461970268
dataset_size: 38586701804
configs:
- config_name: default
data_files:
- split: train
path: data/train-*