metadata
dataset_info:
features:
- name: id
dtype: string
- name: title
dtype: string
- name: context
dtype: string
- name: question
dtype: string
- name: answers
sequence:
- name: text
dtype: string
- name: answer_start
dtype: int32
- name: prompt
dtype: string
- name: ppl_gpt2_med_s15e22_ft_2
dtype: float64
splits:
- name: train
num_bytes: 138319124
num_examples: 77087
download_size: 86665590
dataset_size: 138319124
configs:
- config_name: default
data_files:
- split: train
path: data/train-*