metadata
dataset_info:
features:
- name: language
dtype: int64
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: depths
sequence: int64
- name: sibling_indices
sequence: int64
- name: tree_attention_mask
sequence: int64
splits:
- name: train
num_bytes: 2380160000
num_examples: 160000
- name: test
num_bytes: 169512020
num_examples: 11395
- name: validation
num_bytes: 164305420
num_examples: 11045
download_size: 40720475
dataset_size: 2713977440
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
- split: validation
path: data/validation-*