metadata
dataset_info:
features:
- name: sentence
dtype: string
- name: last_hidden_state
sequence:
sequence: float32
- name: pooler_output
sequence: float32
splits:
- name: train
num_bytes: 19382437136
num_examples: 100000
download_size: 19503523505
dataset_size: 19382437136
configs:
- config_name: default
data_files:
- split: train
path: data/train-*