metadata
dataset_info:
features:
- name: id
dtype: string
- name: input_prompt
dtype: string
- name: correct_answer
dtype: string
- name: choice_logits
dtype: string
- name: best_choice
dtype: string
- name: model_score
dtype: string
splits:
- name: train
num_bytes: 2354349
num_examples: 3270
download_size: 1341788
dataset_size: 2354349
configs:
- config_name: default
data_files:
- split: train
path: data/train-*