metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: depths
sequence: int64
- name: sibling_idxs
sequence: int64
splits:
- name: train
num_bytes: 57731178320
num_examples: 5361365
download_size: 7319941850
dataset_size: 57731178320
configs:
- config_name: default
data_files:
- split: train
path: data/train-*