metadata
dataset_info:
features:
- name: query
dtype: string
- name: context
dtype: string
- name: answer
sequence: string
- name: few_shot_scores
dtype: 'null'
- name: dataset
dtype: string
- name: language
dtype: string
- name: idx
dtype: int64
splits:
- name: train
num_bytes: 4420169
num_examples: 150
download_size: 1835911
dataset_size: 4420169
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
Filtered MultiFieldQA-en dataset, includes only instances where the context lengths is smaller than 10,000 tokens.