metadata
dataset_info:
features:
- name: text_id
dtype: int64
- name: total_log_prob_real
dtype: float64
- name: avg_log_prob_real
dtype: float64
- name: word_probabilities_real
dtype: string
- name: total_log_prob_shifted
dtype: float64
- name: avg_log_prob_shifted
dtype: float64
- name: word_probabilities_shifted
dtype: string
splits:
- name: train
num_bytes: 3343752
num_examples: 282
download_size: 2151248
dataset_size: 3343752
configs:
- config_name: default
data_files:
- split: train
path: data/train-*