metadata
dataset_info:
features:
- name: prompt
dtype: string
- name: gt_is_correct
dtype: bool
- name: gt_token_count
dtype: int64
- name: source_model
dtype: string
- name: original_id
dtype: string
- name: prompt_tokens
dtype: int64
- name: num_anchors
dtype: int64
splits:
- name: train
num_bytes: 12332077
num_examples: 2600
- name: test
num_bytes: 3579171
num_examples: 650
download_size: 5159302
dataset_size: 15911248
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*