metadata
dataset_info:
features:
- name: text_before
dtype: string
- name: text_after
dtype: string
- name: part_type
dtype: string
- name: original_text
dtype: string
- name: input_ids
sequence: int32
- name: labels
sequence: int64
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 381377712
num_examples: 129012
download_size: 120241427
dataset_size: 381377712
configs:
- config_name: default
data_files:
- split: train
path: data/train-*