metadata
dataset_info:
features:
- name: input_ids
sequence: int64
- name: loss_weights
sequence: float32
splits:
- name: train
num_bytes: 430379054112
num_examples: 2188932
- name: validation
num_bytes: 53797283456
num_examples: 273616
- name: test
num_bytes: 53797480072
num_examples: 273617
download_size: 17119735507
dataset_size: 537973817640
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*