metadata
dataset_info:
features:
- name: text
dtype: string
- name: meta
struct:
- name: pile_set_name
dtype: string
- name: sentences_perturbed
dtype: int64
- name: doc_stats
struct:
- name: advantage_score_max
dtype: float64
- name: advantage_score_mean
dtype: float64
- name: advantage_score_min
dtype: float64
- name: advantage_score_std
dtype: float64
- name: advantage_scores
list: float32
- name: baseline_doc_loss
dtype: float32
- name: baseline_truncated
dtype: bool
- name: doc_advantage_score
dtype: float32
- name: length_change_ratio
dtype: float64
- name: net_length_change_chars
dtype: int64
- name: original_sentences_length_chars
dtype: int64
- name: replacement_paraphrases_length_chars
dtype: int64
- name: sentences_perturbed
dtype: int64
- name: target_doc_loss
dtype: float32
- name: target_truncated
dtype: bool
- name: threshold_passing_sentences
dtype: int64
- name: total_doc_length_chars
dtype: int64
- name: total_sentences
dtype: int64
- name: truncated
dtype: bool
splits:
- name: validation
num_bytes: 11326283
num_examples: 4000
download_size: 4772520
dataset_size: 11326283
configs:
- config_name: default
data_files:
- split: validation
path: data/validation-*