metadata
dataset_info:
features:
- name: text
dtype: string
- name: timestamp
dtype: string
- name: url
dtype: string
- name: source
dtype: string
- name: label_identity_attack
dtype: float64
- name: label_insult
dtype: float64
- name: label_obscene
dtype: float64
- name: label_severe_toxicity
dtype: float64
- name: label_threat
dtype: float64
- name: label_toxicity
dtype: float64
- name: perplexity_kenlm
dtype: int64
splits:
- name: train
num_bytes: 116644096770.46587
num_examples: 28815814
download_size: 73370870606
dataset_size: 116644096770.46587
configs:
- config_name: default
data_files:
- split: train
path: data/train-*