metadata
dataset_info:
features:
- name: id
dtype: string
- name: question
dtype: string
- name: correct_answer
dtype: string
- name: distractors
sequence: string
- name: prompt
dtype: string
- name: correct_answer_letter
dtype: string
- name: desired_answer
dtype: string
- name: desired_answer_letter
dtype: string
- name: predicted_answer
dtype: string
- name: predicted_answer_letter
dtype: string
- name: answer_original_text
dtype: string
- name: sandbagging_environment
dtype: bool
- name: prefix
dtype: 'null'
- name: predicting_model_id
dtype: string
splits:
- name: train
num_bytes: 565010
num_examples: 1028
- name: validation
num_bytes: 552167
num_examples: 1000
- name: test
num_bytes: 541906
num_examples: 1000
download_size: 691877
dataset_size: 1659083
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*