metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: labels
sequence: int64
splits:
- name: train
num_bytes: 2894277696
num_examples: 1726896
download_size: 802132482
dataset_size: 2894277696