Skip to content

Scruples

CommonsenseEnglish

The Scruples dataset is a English commonsense resource from Lourie et al. at 2020 comprising 32,766 examples.

About Scruples

Scruples contains 2 datasets: Anecdotes and Dilemmas. Anecdotes contains 32,000 real-life anecdotes about complex ethical situations, with 625,000 ethical judgments extracted from reddit. Dilemmas contains 10,000 ethical dilemmas in the form of paired actions, where the model must identify which one was considered less ethical by crowd workers on Mechanical Turk.

Details

Task
Commonsense
Language
English
Format
JSON
Rows / instances
32,766
Creator
Lourie et al.
Year
2020
Download Paper

Related Commonsense datasets

FAQ