metadata
dataset_info:
features:
- name: id
dtype: string
- name: prompt
dtype: string
- name: answer
dtype: int64
- name: context_length
dtype: int64
- name: token_count
dtype: int64
- name: num_shots
dtype: int64
- name: num_digits
dtype: int64
- name: depth_percent
dtype: int64
splits:
- name: train
num_bytes: 42575600
num_examples: 12600
download_size: 3373964
dataset_size: 42575600
configs:
- config_name: default
data_files:
- split: train
path: data/train-*