metadata
dataset_info:
features:
- name: id
dtype: int64
- name: id_bench
dtype: string
- name: benchmark
dtype: string
- name: prompt
dtype: string
- name: shot_indices
sequence: string
- name: label
dtype: string
- name: model_name
dtype: string
- name: model_answer
dtype: string
- name: parsed_model_answer
dtype: string
splits:
- name: train
num_bytes: 1944109
num_examples: 900
download_size: 100539
dataset_size: 1944109
configs:
- config_name: default
data_files:
- split: train
path: data/train-*