metadata
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
dataset_info:
features:
- name: text1
dtype: string
- name: text2
dtype: string
- name: tgt_input_ids
sequence: int64
- name: tgt_attention_mask
sequence: int64
- name: lm_input_ids
sequence: int64
- name: lm_attention_mask
sequence: int64
- name: atk_input_ids
sequence: int64
- name: atk_attention_mask
sequence: int64
- name: atk_token_type_ids
sequence: int64
splits:
- name: train
num_bytes: 7470489671
num_examples: 1000000
download_size: 557393595
dataset_size: 7470489671