metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: token_type_ids
sequence: int8
- name: attention_mask
sequence: int8
- name: heads
sequence: int64
- name: relns
sequence: int8
splits:
- name: train
num_bytes: 73863703574
num_examples: 9593815
- name: validation
num_bytes: 3887372604
num_examples: 504918
download_size: 16897147862
dataset_size: 77751076178