metadata
dataset_info:
features:
- name: id
dtype: string
- name: prompt
dtype: string
- name: answer
dtype: int64
- name: context_length
dtype: int64
- name: num_tokens
dtype: int64
- name: depth_percent
dtype: int64
splits:
- name: train
num_bytes: 1971935658
num_examples: 77250
download_size: 105915472
dataset_size: 1971935658
configs:
- config_name: default
data_files:
- split: train
path: data/train-*