metadata
license: mit
task_categories:
- question-answering
- text-generation
- text2text-generation
language:
- en
tags:
- natural-questions
- question-answering
- text-generation
- text2text
pretty_name: natural-questions-clean
size_categories:
- 100K<n<1M
configs:
- config_name: raw
data_files:
- split: train
path: raw/train.jsonl
- split: validation
path: raw/validation.jsonl
- config_name: either
data_files:
- split: train
path: either/train.jsonl
- split: validation
path: either/validation.jsonl
default: true
- config_name: long
data_files:
- split: train
path: long/train.jsonl
- split: validation
path: long/validation.jsonl
- config_name: short
data_files:
- split: train
path: short/train.jsonl
- split: validation
path: short/validation.jsonl