metadata
dataset_info:
features:
- name: category
dtype: string
- name: category_id
dtype: int64
- name: prompt
dtype: string
- name: output
dtype: string
- name: scores
struct:
- name: IDENTITY_ATTACK
dtype: float64
- name: INSULT
dtype: float64
- name: PROFANITY
dtype: float64
- name: SEVERE_TOXICITY
dtype: float64
- name: THREAT
dtype: float64
- name: TOXICITY
dtype: float64
- name: caffe_bert_score
list:
- name: label
dtype: string
- name: score
dtype: float64
- name: predict_bert
sequence: string
- name: predict_google
sequence: string
splits:
- name: train
num_bytes: 190683
num_examples: 200
download_size: 101992
dataset_size: 190683
configs:
- config_name: default
data_files:
- split: train
path: data/train-*