metadata
dataset_info:
features:
- name: correct_answer
dtype: string
- name: index
dtype: int64
- name: prompt
dtype: string
- name: sandbagging_environment
dtype: bool
- name: desired_answer
dtype: string
- name: prefix
dtype: string
- name: template
dtype: string
splits:
- name: train
num_bytes: 8287205
num_examples: 11679
- name: validation
num_bytes: 715261
num_examples: 1000
- name: test
num_bytes: 705347
num_examples: 1000
download_size: 4497942
dataset_size: 9707813
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*