metadata
configs:
- config_name: default
data_files:
- split: top_ppl
path: data/top_ppl-*
- split: bottom_ppl
path: data/bottom_ppl-*
dataset_info:
features:
- name: id
dtype: string
- name: title
dtype: string
- name: context
dtype: string
- name: question
dtype: string
- name: answers
struct:
- name: answer_start
sequence: int32
- name: text
sequence: string
- name: prompt
dtype: string
- name: ppl_gpt2-large
dtype: float64
splits:
- name: top_ppl
num_bytes: 42758098
num_examples: 23126
- name: bottom_ppl
num_bytes: 38983057
num_examples: 23126
download_size: 49970093
dataset_size: 81741155