metadata
license: apache-2.0
language:
- en
task_categories:
- summarization
- text2text-generation
tags:
- summarization
- long-document
pretty_name: SQuALITY v1.3
size_categories:
- n<1K
SQuALITY - v1.3
Original paper here
This is v1.3, the 'text' edition .jsonl
files. See description from the original repo:
v1.3 fixes some bugs in v1.2. In v1.2, 10 out of 127 articles (each ~5k-word-long) are missing a few hundreds words each, so summaries may not be fully contained in the article. To fix this issue, we have updated the 10 articles.
contents
again, this is taken from the repo
Each data file ({train/dev/test}.jsonl) is formatted as a JSON lines file. Each row in the data file is a JSON dictionary with the following fields:
- metadata: the Gutenberg story ID, an internal UID, and the Project Gutenberg license
- document: the Gutenberg story
questions: a list of questions and accompanying responses
- question text
- question number: the order in which that question was answered by the writers
- responses: list of worker's response, where each response is a dictionary containing the (anonymized) worker ID, an internal UID, and their response to the question
dataset contents
DatasetDict({
train: Dataset({
features: ['metadata', 'document', 'questions'],
num_rows: 50
})
test: Dataset({
features: ['metadata', 'document', 'questions'],
num_rows: 52
})
validation: Dataset({
features: ['metadata', 'document', 'questions'],
num_rows: 25
})
})