metadata
dataset_info:
features:
- name: input_values
sequence:
sequence:
sequence: float32
- name: attention_mask
sequence:
sequence: int32
- name: labels
dtype: int64
splits:
- name: train
num_bytes: 10873164220
num_examples: 29939
download_size: 9851610543
dataset_size: 10873164220