metadata
license: mit
dataset_info:
features:
- name: text
dtype: string
- name: log_prob
dtype: float64
- name: perplexity
dtype: float64
- name: num_tokens
dtype: int64
splits:
- name: train
num_bytes: 413246
num_examples: 5000
- name: valid
num_bytes: 41314
num_examples: 500
- name: test
num_bytes: 41391
num_examples: 500
- name: unlabeled
num_bytes: 831779
num_examples: 10000
download_size: 537280
dataset_size: 1327730
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: valid
path: data/valid-*
- split: test
path: data/test-*
- split: unlabeled
path: data/unlabeled-*