metadata
dataset_info:
config_name: '{''do_sample''=False, ''beams''=1}'
features:
- name: id
dtype: string
- name: prediction
dtype: string
- name: winogrande_accuracy
dtype: bool
splits:
- name: train
num_bytes: 121601
num_examples: 1267
download_size: 31928
dataset_size: 121601
configs:
- config_name: '{''do_sample''=False, ''beams''=1}'
data_files:
- split: train
path: '{''do_sample''=False, ''beams''=1}/train-*'