metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: generation
dtype: string
- name: generation_method
dtype: string
- name: group
dtype: string
- name: prompt_label
dtype: int64
- name: roberta_prediction
dtype: float64
- name: refusal
dtype: string
splits:
- name: train
num_bytes: 60607410.45853573
num_examples: 62737
- name: eval
num_bytes: 4830276.428466115
num_examples: 5000
download_size: 25319599
dataset_size: 65437686.88700184
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: eval
path: data/eval-*