metadata
language:
- en
dataset_info:
features:
- name: question
dtype: string
- name: answer
dtype: string
- name: negative_1
dtype: string
- name: negative_2
dtype: string
- name: negative_3
dtype: string
- name: negative_4
dtype: string
- name: negative_5
dtype: string
- name: negative_6
dtype: string
- name: negative_7
dtype: string
- name: negative_8
dtype: string
- name: negative_9
dtype: string
- name: negative_10
dtype: string
splits:
- name: train
num_bytes: 2811249248
num_examples: 1000000
download_size: 1897316597
dataset_size: 2811249248
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
Warning: Do not use this for training - the answer is listed as one of the negatives.
It is only recommended to use this for evaluation of CrossEncoder models.